Introduction of a single carboxylic acid converts the cyclic oligomeric depsipeptide ent-verticilide from a ryanodine receptor 2 (RyR2) inhibitor to RyR2 activator

Abstract

Cyclic oligomeric depsipeptides represent a distinct structural class of naturally occurring compounds known for their wide-ranging biological activities. We previously reported that the unnatural form of verticilide (ent-verticilide) inhibits cardiac ryanodine receptor 2 (RyR2) and exhibits antiarrhythmic effects in mice, but its mechanism of action on the RyR2 channel is not known. Here, we collected single-channel recordings in artificial lipid bilayers to elucidate the mechanism of RyR2 modulation by ent-verticilide and its polar side chain analog activert. ent-Verticilide reduced RyR2 activity by increasing the RyR2 mean closed time without changing the RyR2 mean open time, suggesting that ent-verticilide functions as a closed-channel stabilizer. ent-Verticilide exhibited partial inhibition on RyR2 single channels with an IC₅₀ of ∼0.2 μM and a maximal inhibitory efficacy of ∼23%. To explore the effect of a charged residue on ent-verticilide-RyR2 binding, we introduced a terminal carboxylic acid on a single pentyl side chain. The resulting compound lost its inhibitory activity, increased RyR1 and RyR2 single-channel activity, and increased Ca spark frequency. Thus, we named this analog activert (1). Single-channel analysis showed that activert shortened mean closed time without changing mean open time, indicating closed-channel destabilization. Compared with ent-verticilide, activert was ∼100-fold less potent (EC₅₀ ∼ 30 μM) on RyR2 and had low membrane permeability. RyR2 activation was confirmed by [³H]-ryanodine binding and Ca spark assays. Although poor membrane permeability represents an obstacle for therapeutic development, activert serves as a proof-of-concept partial RyR2 activator and a promising scaffold for future structure–activity optimization.

Significance Statement

This study reveals the dual modulatory potential of cyclooligomeric depsipeptides on ryanodine receptor 2, with ent-verticilide acting as a closed-channel stabilizer and its analog, activert, functioning as a closed-channel destabilizer. By leveraging nonnatural enantiomers and rational scaffold modifications, we highlight an underexplored approach to uncover important structure-activity relationships, advancing the development of novel ryanodine receptor 2-targeted therapeutics with potential applications in cardiac arrhythmia management.

1. Introduction

Calcium (Ca) movement across the sarcolemma and the sarcoplasmic reticulum (SR) membrane of cardiomyocytes plays a central role in cardiac excitation-contraction coupling.¹ Depolarization of sarcolemma leads to the opening of voltage-gated L-type Ca channels, which promote Ca entry. The ensuing Ca flux triggers Ca release from the SR via ryanodine receptor 2 (RyR2) and promotes cardiac contraction, collectively termed “excitation-contraction” coupling. With membrane repolarization, L-type Ca and RyR2 channels close and SR Ca release terminates. RyR2 hyperactivity either due to gain-of-function mutations or structural heart disease causes SR Ca leak and is associated with lethal cardiac arrhythmias such as catecholaminergic polymorphic ventricular tachycardia (CPVT).2, 3, 4 Consequently, RyR2 inhibitors may be viable therapeutics for preventing cardiac arrhythmias. Several agents that inhibit RyR2, including flecainide, propafenone, and dantrolene, have been shown to reduce spontaneous diastolic SR Ca release events and are antiarrhythmic in mice.5, 6, 7, 8, 9, 10 However, those agents often cause serious side effects and have proarrhythmic liability in structural heart disease.11, 12, 13 Therefore, there is a need for more selective and potent RyR2 inhibitors.

We recently discovered that the unnatural enantiomer of (−)-verticilide (ent-(+)-verticilide) selectively inhibits RyR2 in ryanodine binding studies and prevents arrhythmia in CPVT mice.14, 15, 16 However, the exact mechanism of ent-verticilide inhibition of RyR2 Ca release channels remains unknown.

Many cyclic oligomeric depsipeptides (CODs) are classified as “Beyond Rule of 5” and ent-verticilide’s oligomeric nature leads one to question whether the pharmacophore for RyR2 regulation may not involve the entire structure.^17,18 We have developed ring-size analogs of ent-verticilide ranging from 6 to 36 where the polar scaffold is altered by merely varying the depsipeptide chain as well as analogs with modifications of the polar backbone.19, 20, 21 In this study, we explore the structure-activity relationship of the aliphatic side chain by introducing a carboxylic acid to one pentyl side chain of ent-verticilide without change to the remaining structure with the goal of increasing solubility. Interestingly, the new analog exhibited partial RyR2 activation (hence we named it activert (1)). Importantly, partial RyR2 activators could have value in arrhythmias caused by RyR2-loss-of-function mutations, a genetic sudden death syndrome characterized by high risk of ventricular arrhythmias—the RyR2 Ca release deficiency syndrome (CRDS).^22,23 A partial RyR2 activator could not only be an enabling tool compound to study disease mechanisms but also serve as a lead for therapeutic drug development. Here, we investigated the direct effect of ent-verticilide and its carboxylic acid side chain analog activert on RyR2 activity using single-channel recording of RyR2 in artificial lipid bilayers and ryanodine binding. We also measured activert actions on RyR2 Ca release using the Ca spark assay and assessed activert activity on RyR1 isolated from skeletal muscles using single channel and [³H] ryanodine binding assay.

2. Materials and methods

2.1. Drugs, chemicals, and reagents

All chemicals and reagents were purchased from Sigma-Aldrich unless otherwise stated. The synthesis of ent-verticilide has been described elsewhere.^15,24 Details of synthesis, purification, and characterization of activert, including NMR spectra, are provided in Supplemental Figs. 1–8 and outlined in Fig. 1. In brief, the oligomeric substructure of activert (1) provides an opportunity for convergency in its synthesis, but the unique didepsipeptide residue must be prepared separately and, preferably, incorporated near the synthesis end. As a result, the preparation of activert was modeled after an analog used previously to incorporate a fluorophore at one α-oxy amide side chain.²⁵ The unique didepsipeptide 4 was prepared from L-(−)-malic acid. Conversion of the α-hydroxy acid first to its para-methoxy benzyl ester, and then alkylation of the remaining acid using benzyl bromide under basic conditions led to alcohol 2 in 50% yield over 2 steps. Steglich esterification with Boc-N-Me-D-Ala proceeded uneventfully to give the fully protected didepsipeptide 3 in 88% yield. Removal of the N-terminal protecting group was accomplished selectively with the use of HCl in ethyl acetate, and the disubstituted amine that resulted was coupled with the hexadepsipeptide described previously.²⁵ The efficiency of this coupling was noted to be sensitive to the coupling procedure, with yields for this step ranging from 25% to 83%. An “activation” period involving only the acid and bromo(tripyrrolidin-1-yl)phosphanium hexafluorophosphate proved helpful. The octadepsipeptide needed selective deprotection at the N- and a single C-terminus. This was accomplished by its treatment with 10% trifluoroacetic acid in dichloromethane. The amino acid so produced (6) was treated with benzotriazol-1-yloxy(tripyrrolidin-1-yl)phosphanium hexafluorophosphate to effect the desired macrocyclization, delivering 7 in 32% yield over 2 steps. Deprotection of the benzyl ester was readily accomplished using standard hydrogenolysis conditions in ethyl acetate, furnishing activert (1) in 82% yield.

Fig. 1.

Synthesis of activert (1). See text and Supplemental Material for more details.

2.2. [³H]-Ryanodine ligand binding assays

[³H]-Ryanodine, [9,21-3H(N)] (56 Ci/mmol) was purchased from PerkinElmer. Porcine cardiac and skeletal SR membrane (3 mg/mL and 1 mg/mL, respectively) was incubated with a range of indicated activert concentrations in a solution containing 20 mM piperazine-N,N[prime]-bis(2-ethanesulfonic acid) (pH 7.0), 150 mM KCl, 5 mM glutathione, 2 mM dithiothreitol, 1 μg/mL aprotinin/leupeptin, 0.1 mg/mL bovine serum albumin, 5 mM ATP, 0.1 μM calmodulin, and 0.1 μM free Ca with 1 mM EGTA as a Ca chelator (as determined by MaxChelator) for 3 hours at 37 °C. The concentrations of [³H]-ryanodine were 7.5 nM and 10 nM for cardiac and skeletal SR, respectively. Nonspecific [³H]-ryanodine binding to SR was assessed by addition of 15 μM nonradioactive ryanodine. Maximal [³H]-ryanodine binding was determined by addition of 5 mM adenylyl-imidodiphosphate supplemented with 20 mM caffeine. After a 3-hour incubation at 37 °C, binding of [³H]-ryanodine was determined by filtration through grade GF/B glass microfiber filters (Brandel Inc) using a 96-well Brandel Harvester. In 4 mL of Ecolite scintillation mixture (MP Biomedicals). The [³H]-ryanodine retained on the filter was measured using a Beckman LS6000 scintillation counter.

2.3. Ca spark assays

The use of animals was approved by the Institutional Animal Care and Use Committee of Vanderbilt University (Protocol #M1900057-01). Ventricular cardiomyocytes were isolated from 12 to 18-week-old wild-type C57BL/6J mice (see figure legends for biological sex) as described previously.²⁶ Isolated ventricular myocytes were plated onto laminin-coated glass cover slips and allowed to attach for 10 minutes at room temperature (20–24 °C). The attached cells were permeabilized with saponin (40 μg/mL) for 40 seconds and then bathed for 15 minutes in a freshly made internal solution (pH 7.2) containing (in mM): K-aspartate (120), KCl (15), K₂HPO₄ (5), MgCl₂ (5.6), HEPES (10), dextran (4% w/v), magnesium ATP (MgATP) (5), phosphocreatine-Na₂ (10), creatine phosphokinase (10 U/mL), reduced L-glutathione (10), EGTA (0.5), CaCl₂ (0.12), and Fluo-4 (0.03). The internal solution contained DMSO at a concentration of 0.1% (v/v) as control or activert in DMSO. Confocal imaging was performed as previously described.²⁷ Sparks were recorded in cells that had characteristic brick-shape morphology. SR Ca image analysis and all the associated parameters were performed and derived using SparkMaster 2.²⁸

2.4. Single-channel recording

SR vesicles containing RyR1 and RyR2 were isolated from longissimus dorsi porcine skeletal muscles and porcine hearts, respectively, and incorporated in artificial bilayer membranes as previously described for sheep RyR2.²⁹ Lipid bilayers were formed across an aperture with diameter 150 to 250 mm of a delrin cup using a lipid mixture of phosphatidylethanolamine, phosphatidylserine, and phosphatidylcholine (5:3:2 wt/wt, Avanti Polar Lipids) in n-decane. During the SR vesicle fusion period, the cis (cytoplasmic) chamber contained 250 mM Cs⁺ (230 mM CH₃SO₃Cs, 20 mM CsCl) + 1 mM CaCl₂ and the trans (luminal) chamber contained 50 mM Cs⁺ (30 mM CH₃SO₃Cs, 20 mM CsCl) + 1 mM CaCl₂. When ion channels were detected in the bilayer, the trans Cs⁺ was raised to 250 mM by aliquot addition of 4 M CH₃SO₃Cs. During experiments, RyR1 and RyR2 channels were exposed to drugs via either aliquot addition or local perfusion to the cis bath as described by O’Neill et al.³⁰ ent-Verticilide and activert were dissolved in DMSO, ensuring that the final DMSO concentration in the cis bath remained below 0.1% (v/v). All solutions were buffered using 10 mM N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid (ICN Biomedicals) and titrated to pH 7.4 using CsOH. Free Ca of 0.1 μM was generated from 1 mM CaCl₂ and 4.5 mM 1,2-bis(oaminophenoxy) ethane-N,N, N′,N′-tetraacetic acid (obtained from Invitrogen) and this was validated using a Ca electrode (Radiometer).

2.5. Parallel artificial membrane permeability assay

Details of the experimental procedure for parallel artificial membrane permeability assay (PAMPA) assay are in the Supplemental Material. In brief, PAMPA was conducted using 96-well donor plates with 0.45 μm hydrophobic Immobilon-P membranes and 96-well Teflon acceptor plates. Stock solutions of compounds (1–4 mg in 100% DMSO) were diluted to 500 μM in PBS (pH 7.4) containing D-α-tocopherol polyethylene glycol 1000 succinate. Membranes were pretreated with 1% lecithin in n-dodecane. Donor solutions (150 μL) were added to the donor wells, and acceptor wells contained 300 μL of PBS. Plates were incubated at room temperature for 14 to 18 hours in a humidified chamber to prevent evaporation. After incubation, donor and acceptor solutions were collected in glass vials for liquid chromatography-mass spectrometry analysis (Agilent 6130 single quadrupole system, Eclipse XD-C18 column). PAMPA permeability (P_app), percent recovery, and percent diffusion were calculated using standard ratio-based formulas (Supplemental eqs. 1–3).

2.6. Polarized human colorectal adenocarcinoma-derived cells assay

Full experimental details for the polarized human colorectal adenocarcinoma-derived cells (Caco-2) assay are included in Supplemental Material. To summarize, bidirectional permeability of compounds was evaluated using differentiated Caco-2 monolayers (passages 55–65) cultured on 0.4 μm Transwell inserts for 21 days. Inserts with transepithelial/endothelial electrical resistance >1000 Ω·cm² were used. Compounds (10 μM in Hanks’ balanced salt solution with <1% DMSO) were applied to apical or basal compartments for apical-to-basal (A-B) and basal-to-apical (B-A) transport studies. Samples were collected at 0 and 2 hours and analyzed by liquid chromatography-mass spectrometry to determine apparent permeability (P_app), efflux ratio, and compound recovery. Monolayer integrity was confirmed postassay using lucifer yellow paracellular flux (≤0.5% considered intact).

2.7. Single-channel data acquisition and analysis

Experiments were carried out at room temperature (23 ± 2 °C) and at physiological Ca level during end-diastole (pCa 7 or 0.1 μM Ca) plus 2 mM ATP unless otherwise stated. ATP, an endogenous coagonist of RYR2,³¹ was added to the cytosolic side to activate RyR2 because at pCa 7, RyR2 open probability (P_o) is too low (P_o ∼0.00001–0.00004) for accurate single-channel measurement.³² Similar conditions were also applied for RyR1. Electric potentials are expressed using standard physiological convention (ie, cytoplasm relative to SR lumen at virtual ground). Control of the bilayer potential and recording of unitary currents was done using an Axopatch 200B amplifier (Axon Instruments/Molecular Devices). Channel currents were digitized at 20 kHz and low pass-filtered at 5 kHz. Before analysis the current signal was redigitized at 5 kHz and low pass-filtered at 1 kHz. Individual readings of P_o were derived from 30 to 60 seconds of RyR1 and RyR2 recording. Single channel P_o, mean open time, and mean closed time were measured using a threshold discriminator at 50% of channel maximal amplitude. Open probabilities from multiple-channel recordings were calculated from the ratio of the mean current and unitary current.

Dwell-time frequency histograms of channel openings were exclusively obtained from single-channel recordings and were displayed as square root of normalized frequency (counts/total numbers of events). These were plotted using the “log-bin” method suggested by Sigworth and Sine,³³ which displays exponential values as peaked distributions centered around their exponential time constant. Sampling bins were equally spaced on a log scale with 7 bins per decade. Open and closed dwell time distributions were fitted by 2 exponential components using a Matlab script.

2.8. Statistical analysis

Statistics were carried out in GraphPad Prism (v9.5.0) or R as detailed in the figure legends. For single-channel data, normality was assessed using the Kolmogorov-Smirnov test with the Dallal-Wilkinson-Lillie method to calculate the P value (KS distance). For groups that passed the normality test, Student’s t test was used to evaluate significant differences. Nonparametric tests were applied for comparisons involving nonnormally distributed data. For Ca spark data, normality was examined using the Shapiro-Wilk test and data were transformed, if necessary, before testing with a mixed-effects hierarchical clustering model to account for variability between animals (detailed results of statistical tests can be found in Supplemental Table 1). Results are presented as mean ± SD. All statistical tests were 2-sided, and significance was defined as a P value < .05. Concentration-response curves for RyR2 single channel, [³H]-ryanodine ligand binding assay, and Ca spark assay were generated using nonlinear regression model with the equation: $Y=Bottom+\frac{Top-Bottom}{1+{\left(\frac{A}{X}\right)}^{HillSlope}}$ (A = IC₅₀ or EC₅₀, X: drug concentrations), using least square criteria, no weighting, and no constraints on the parameters.

3. Results

3.1. Effect of ent-verticilide on RyR2 single-channel recordings

The effect of ent-verticilide (structure shown in Fig. 2A) on RyR2 single channels was examined by measuring the ratio of RyR2 P_o during 30- to 60-second drug exposure to the P_o of 30 to 60 second periods of vehicle solution. Consistent with an inhibitory action, ent-verticilide shifted the current amplitude histogram probability toward I = 0 with no observable substates (Fig. 2, B and D). We next evaluated the concentration dependence using 8 concentrations of ent-verticilide (0.01, 0.03, 0.1, 0.3, 1, 3, 6, and 10 μM) on RyR2 P_o. ent-Verticilide inhibited RyR2 channels in a concentration-dependent manner, with an estimated IC₅₀ of ∼0.2 μM (95% confidence interval [CI], 0.026–1.40 μM) with incomplete maximal inhibitory efficacy (I_max ∼ 23%, 95% CI, 15%–33%) (Fig. 2C). The partial inhibitory effects of ent-verticilide observed in RyR2 single-channel recordings align with its effects in [³H]ryanodine binding and spontaneous Ca release assays described in earlier studies.¹⁵

Fig. 2.

ent-Verticilide inhibits RyR2 single channel. (A) Structure of ent-verticilide and activert. (B) Representative traces of single-channel recording in the presence of vehicle (DMSO) or 3 μM ent-verticilide. The bilayer voltage was clamped at +40 mV. The cis bath contained 0.1 μM Ca (pCa 7) and 2 mM ATP. Channel openings (O) were indicated as current jumps from close (C) baseline. (C) Concentration-response curve with increasing concentrations of ent-verticilide in μM and the number of data points for each concentration (0.01 [n = 13]; 0.03 [n = 12], 0.1 [n = 12], 0.3 [n = 6], 1 [n = 16], 3 [n = 14], 6 [n = 7], 10 [n = 6]). The relative P_o was calculated as the ratio of RyR2 P_o in the presence of ent-verticilide and RyR2 P_o in the presence of vehicle on the same RyR2 channel. Nonlinear regression with 4-parameter fit using least square criteria produced an IC₅₀ ∼ 0.2 μM (95% CI, 0.026–1.4 μM) and I_max ∼ 23%, (95% CI, 15%–33%). Data are shown as mean ± SD. ∗P = .0054, .0147, .0141, and .0179 by 1-sample Student’s t test against null hypothesis of 1 (not correcting for multiple comparisons) for concentrations at 1, 3, 6, and 10 μM, respectively. (D) Current amplitude histogram of vehicle and 3 μM ent-verticilide from (B). Probability was calculated as count divided by total number of events.

3.2. Effect of ent-verticilide on RyR2 dwell time

To further investigate the mechanism underlying ent-verticilide RyR2 inhibition, we constructed dwell-time histograms for channel open and closed events. In the absence of ent-verticilide, open and closed dwell times exhibited peaks that were resolved by fitting 2 exponential components. In the presence of 3 μM ent-verticilide, no changes were observed for the open dwell time (Fig. 3A; representative of 5 experiments detailed in Supplemental Table 2). However, analysis of the closed dwell time revealed 2 distinct exponential components. ent-Verticilide increased the longer time-constant of the closed dwell time, whereas the shorter time constant was not significantly changed (Fig. 3B; Supplemental Table 2). In addition, ent-verticilide reduced RyR2 P_o by prolonging the mean channel closed time without altering the mean channel open time (Fig. 3, C and D). Collectively, these findings suggest that ent-verticilide inhibits RyR2 by stabilizing its closed state.

Fig. 3.

ent-Verticilide effects on dwell times. (A) Open and close (B) dwell time histogram of vehicle (■) and 3 μM ent-verticilide (▲). These were plotted using the log-bin method suggested by Sigworth and Sine that displays exponential values as peaked distributions centered around their exponential time constant. Histograms were compiled from 60-second recordings. Statistical analysis of open (C) or close (D) dwell-time showing relative changes in mean dwell time over a range of ent-verticilide concentrations versus vehicle (DMSO). Data were log-transformed. Asterisks indicate significant difference from 0 (dotted line) using 1-sample Student’s t test against null hypothesis of 0, with P value (left to right for open and close dwell time for each concentration and not correcting for multiple comparisons) .8587, .2147, .4105, .6064, .0447, .0490, .0356, and .0358. Data are shown as mean ± SD. Normality was examined before statistical testing.

3.3. Activert, a single residue analog of ent-verticilide, activates RyR2

Because ent-verticilide represents the first unnatural COD to selectively block RyR2,^14,15 we leveraged its scaffold to investigate the structure-activity relationship (SAR) of CODs in RyR2 inhibition. We previously investigated ring-size analogs of ent-verticilide ranging from 6 to 36 (ent-verticilide = 24),²¹ which reduced inhibitory potency, and the role of the polar backbone of ent-verticilide by substituting the ester functionality for N-H and N-Me amide,²⁰ which also reduced inhibitory activity. These effects are independent of SAR for passive permeability.^27,34 Here, we modified a single pentyl chain of ent-verticilide to a carboxylic acid (1, “activert” Fig. 1). We first examined the effect of activert on RyR2-mediated Ca release in a Ca spark assay carried out in permeabilized cardiomyocytes isolated from C57BL/6J mice. Ca sparks serve as an index of RyR2 activity and have been used extensively to quantify RyR2 function.³⁵ Figure 4 shows representative x-t fluorescence traces of Ca sparks (bright spots) in permeabilized cardiomyocytes treated with either vehicle (DMSO) or 25 μM activert for 15 minutes before recording. There was an increase in the frequency of Ca sparks in cells treated with activert (Fig. 4B), without changes in the size (amplitude or width) of the average spark (Fig. 4, C and D). The SR Ca load, as measured by the amplitude of caffeine-induced Ca transient, was reduced by ∼10% in cells treated with activert (Fig. 4E), which is consistent with an RyR2 activator. The overall RyR2-mediated SR Ca leak was elevated in permeabilized cardiomyocytes treated with activert (Fig. 4F). In addition, activert modestly increased the average duration of the Ca spark (Fig. 4, G and A inset).

Fig. 4.

Ca spark properties in permeabilized cardiomyocytes treated with vehicle or activert. (A) Representative x-t traces showing Ca sparks (bright spots). (B) Activert increased the frequency of Ca sparks, however, the (C) amplitude and (D) remained unchanged. (E) Cells treated with activert had a smaller SR Ca load, as measured by the amplitude of the caffeine-induced Ca transient. (F) The overall Ca leak, normalized to the SR load and relative to vehicle, was greater in cells treated with activert. (G) Ca sparks lasted longer in cells treated with activert. N = 99 (DMSO) and 91 (activert) cells from 4 mice (3 male and 1 female). Statistical comparisons made using hierarchical clustering mixed-effects model. P values are indicated in each panel. FDHM, full duration at half maximum; FWHM, full width at half maximum.

We next examined the concentration-dependent effect on Ca sparks with activert (Fig. 5A). At the highest concentration tested (100 μM), we observed protracted spark events commonly referred to as “embers” (Fig. 5B, white arrowheads) that failed to terminate within the typical time frame. The distribution of spark duration was prolonged with 100 μM activert compared with DMSO control (Fig. 5C). This was similar to observations made with the known RyR2 activator caffeine,³⁶ where recordings at 1 mM caffeine produced many long-lasting sparks (Fig. 5D).

Fig. 5.

Ca spark properties in permeabilized cardiomyocytes treated with activert or caffeine. (A) Activert produced a dose-dependent increase in Ca spark frequency. Data presented as mean spark frequency from 7, 3, 2, 3, 5, and 1 mice (5 male and 2 female), respectively. (B) Representative x-t traces showing Ca sparks (bright spots) in cells treated with 100 μM activert or 1 mM caffeine. Prolonged sparks are indicated by white triangles. (C) The cumulative distribution of Ca spark full duration at half max relative to DMSO control. N = 1923 and 3445 sparks for DMSO and activert, respectively. (D) Caffeine dose-dependent response for Ca spark frequency. Data presented as mean ± SD from 22, 17, 37, 26, 24, and 20 cells, respectively, from 1 female mouse. FDHM, full duration at half maximum.

Consequently, the methyl ester of 1 (Supplemental Fig. 9A) lacked either RyR2 inhibitory or activating properties (Supplemental Fig. 9B). Incubation with both activert and ent-verticilide showed overall partial inhibition of RyR2 by ∼50%, suggesting a type of competitive antagonism (Supplemental Fig. 9C).

We next performed a [³H]-ryanodine binding assay using SR vesicles isolated from porcine heart. In this assay, the presence of an activator increases [³H]-ryanodine binding to RyR2 because ryanodine binds to the RyR2 open state.³⁷ Compared with vehicle control, activert increased [³H]-ryanodine binding with a half maximal efficacy EC₅₀ of ∼46 μM (95% CI, 41.2–51.6 μM) and incomplete maximal activation (E_max ∼ 29%; 95% CI, 26.6%–31.2%) (Fig. 6A). These findings also indicate that activert directly activates RyR2.

Fig. 6.

Activity of activert in RyR2 single channels. (A) [³H]-Ryanodine binding to porcine RyR2 in the presence of activert. RyR2 activation was measured by the fraction of [³H]-ryanodine binding to RyR2 in the presence of activert normalized to vehicle. Concentration-response curve using nonlinear regression with 4-parameter fit produced EC₅₀ ∼ 46 μM (95% CI, 41.2–51.6 μM) and E_max ∼ 29% (95% CI, 26.6%–31.2%). Data are shown as mean ± SD. ∗∗∗P = .0006 and .0002 by 1-sample Student’s t test for concentrations at 60 and 100 μM, respectively. (B) Representative traces of single-channel recording in the presence of vehicle (DMSO) or 50 μM activert. The bilayer voltage was clamped at +40 mV. The cis bath contained 0.1 μM Ca (pCa 7) and 2 mM ATP. Channel openings (O) were indicated as current jumps from close (C) baseline. (C) Concentration-response curve with increasing concentrations of activert in μM and number of points for each concentration (1 [n = 7], 3 [n = 10], 10 [n = 11], 25 [n = 16], 50 [n = 15], 100 [n = 17]). Nonlinear regression with 4-parameter fit using least square criteria produced an IC₅₀ ∼ 32.2 μM (95% CI, 1.4–63.1 μM) and E_max ∼ 34.0%, (95% CI, 1.0%–63.0%). Data are shown as mean ± SD. ∗P = .0369 and .0483 by 1-sample student’s t test against the null hypothesis of 1 for concentrations at 50 and 100 μM (not corrected for multiple comparisons), respectively. (D) Current amplitude histogram of vehicle and 50 μM activert from (B).

3.4. Activity of activert in RyR2 single-channel recordings

As activert is the first ent-verticilide analog that activated RyR2, we next investigated its mechanism of RyR2 activation using single-channel recordings. First, the effect of increasing cytoplasmic activert concentration on RyR2 channel was studied. Activert exhibited concentration-dependent activation in the single-channel experiment (Fig. 6C) with an EC₅₀ of ∼32.2 μM (95% CI, 1.4–63.1 μM) and partial activation (E_max ∼ 34.0%; 95% CI, 1.0%–63.0%). This result aligns with the [³H]-ryanodine binding results, which demonstrated comparable EC₅₀ and E_max value. Furthermore, in the presence of activert, the current amplitude histogram probability shifted toward the unitary current (I ∼ 12 pA) with no detectable substate (Fig. 6, B and D).

Figure 7 shows dwell-time histograms of channel openings (Fig. 7A) and closures (Fig. 7B) in the absence and presence of activert (all histograms could be fitted by the sum of 2 exponentials). Under control conditions, the open and closed dwell times displayed peaks that were resolved by fitting 2 exponential components (Supplemental Table 3). Activert decreased the slow time-constant of the closed time distribution, but no significant change was observed for the fast time-constant (Supplemental Table 3). In addition, no change was observed for the open dwell time in the presence of activert compared with control. Further analysis on the mean open and closed time revealed that at high concentrations (≥50 μM), activert decreased mean closed duration without altering mean open duration (Fig. 7, C and D). Taken together, the single-channel experiment showed that activert promotes RyR2 activity by destabilizing the RyR2 closed state and thereby increasing RyR2 open frequency. These results are consistent with the increased spark frequency seen in the Ca spark assay with activert (Fig. 4B).

Fig. 7.

Activert activates RyR2 by destabilizing the RyR2 channel closed state. (A) Open and close B) dwell time histogram of vehicle (●) and 50 μM activert (■). These were plotted using the log-bin method suggested by Sigworth and Sine that displays exponential values as peaked distributions centered around their exponential time constant. Histograms were compiled from 60-second recordings. Statistical analysis of open (C) or close (D) dwell-time showing relative changes in mean dwell time over a range of activert concentrations versus vehicle (DMSO). Data were log-transformed. Asterisks indicate significant difference from 0 (dotted line) using 1-sample Student’s t test, with P value for each concentration (left to right for open and close dwell time, not correcting for multiple comparisons) .5322, .0969, .8181, .1336, .6141, .0891, .4726, .4959, .8143, .6765, .0020, and .0019. Data are shown as mean ± SD. Normality was examined before statistical testing.

3.5. Activert partially activated RyR1 from porcine skeletal muscle

To evaluate the potential effects of activert on RyR1, we performed [³H]-ryanodine binding assay and single-channel experiment in RyR1 isolated from longissimus dorsi porcine skeletal muscle. Activert enhanced binding of radiolabeled ryanodine, exhibiting an EC₅₀ ∼ 9 μM (95% CI, 3.1–27.2 μM) and E_max ∼ 17% (95% CI, 13.5%–21.9%) (Fig. 8A). In parallel, single-channel analysis revealed that activert increased RyR1 P_o with an EC₅₀ ∼ 10.5 μM (95% CI, 1.6–28.7 μM) and E_max ∼ 14.4% (95% CI, 6.9%–15.7%) (Fig. 8B). Collectively, these findings indicated that activert acted as a weak activator of RyR1 yet demonstrated greater potency toward RyR1 than previously observed for its action in RyR2.

Fig. 8.

Activity of activert in RyR1 isolated from porcine longissimus dorsi skeletal muscle. (A) [³H]-Ryanodine binding to porcine RyR1 in the presence of activert. RyR1 activation was measured by the fraction of [³H]-ryanodine binding to RyR1 in the presence of activert normalized to vehicle. Concentration-response curve using nonlinear regression with 4-parameter fit produced EC₅₀ ∼ 9 μM (95% CI, 3.1–27.2 μM) and E_max ∼ 17% (95% CI, 13.5%–21.9%). Data are shown as mean ± SD. ∗∗P = .0018 and ∗P = .0108 by 1-sample Student’s t test against the null hypothesis of 1 (not correcting for multiple comparisons) for concentrations at 60 and 100 μM, respectively. (B) Activert activated RyR1 single channel. Concentration-response curve using nonlinear regression with 4-parameter fit yielded EC₅₀ ∼ 10.5 μM (95% CI, 1.6–28.7 μM) and E_max ∼ 14.4% (95% CI, 6.9%–15.7%). ∗P = .0395, .0361, and .0241 by 1-sample Student’s t test for concentration at 50, 75, and 100 μM, respectively.

3.6. Membrane permeability of ent-verticilide and activert

Because RyR2 is an intracellular receptor, compounds must cross the cell membrane to exert their RyR2-derived pharmacological effects. Consequently, membrane permeability can directly influence target engagement and therapeutic potential. To begin assessing the potential of ent-verticilide and activert as therapeutic agents, we quantified their membrane permeability using the SwissADME online prediction tool,³⁸ and the PAMPA and Caco-2 experimental assays (Table 1).³⁹ Based on the predicted and experimental membrane permeability data, ent-verticilide can be classified as a moderately permeable drug, whereas activert exhibits low membrane permeability.

Table 1.

Predicted and experimental membrane permeability of ent-verticilide and activert

Assay	Parameter	ent-Verticilide	Activert
SwissADME	TPSA (Å2)	186	224
	AlogP	6.31	5.20
Caco-2	Papp (A-B) (10−6 cm/s)	1.2	0.4
	Papp (B-A) (10−6 cm/s)	1.5	1.1
	Papp (10−6 cm/s)	Moderate	Low
	ER	1.30	2.73
	%R (A-B)	29	101
	%R (B-A)	86	80
	Permeability	Moderate	Low
PAMPA	Papp (10−6 cm/s)	6.92 ± 0.66	2.17 ± 0.42
	%R	94	92
	% Diffusion	26	7
	Permeability	Moderate	Low

Membrane permeability of ent-verticilide and activert was assessed using the SwissADME online prediction tool, the parallel artificial membrane permeability assay (PAMPA) and the polarized human colorectal adenocarcinoma-derived cells (Caco-2) assay. See Supplemental Material for details of the experimental procedure. TPSA, topological polar surface area using SwissADME (given as Ų); AlogP, calculated LogP using SwissADME; P_app, experimental apparent permeability (given as ×10⁻⁶ cm/s), which corresponds to the proportion of compound that crosses the cell monolayer at each time point divided by the product of the cell monolayer surface area and the initial concentration of the compound in the donor side at time zero; A-B and B-A refer to Apical-to-Basal and Basal-to-Apical direction of each compound across Caco-2 cell monolayer; %R, % recovery (amount of analyte recovered in PAMPA/Caco-2 assay); Permeability, experimental permeability rate; ER, efflux ratio, ER was used a general measure to evaluate the involvement of passive and active transport processes using the following equation: $ER=\frac{PappB-A}{PappA-B}$, where P_{app B-A} and P_{app A-B} refers to the permeability of each compound in the B-A and A-B directions, respectively; P_{app A-B} values in the range of 1–10 × 10⁻⁶ cm/s are classified as moderate and low-permeability drugs are characterized by a P_{app A-B} below 1 × 10⁻⁶ cm/s³⁹; % Diffusion, amount diffused in PAMPA assay. For PAMPA assay, low-permeability cut-off is P_app < 2 × 10⁻⁶ cm/s, moderate permeability P_app 2-10 × 10⁻⁶ cm/s, and high permeability > 10 × 10⁻⁶ cm/s (ref in page 70, supplement information of ref²⁷). The PAMPA data shown is an average of 45 experiments (ent-verticilide) or 20 experiments (activert).

4. Discussion

4.1. Mechanism of ent-verticilide inhibition on RyR2 channels

This study provides the first detailed demonstration of the mechanism by which ent-verticilide inhibits RyR2 channels using single-channel recording. The action of ent-verticilide on cardiac RyR2 shares some similarities to other closed-channel stabilizers including dantrolene and a group of benzothiazepine derivatives known as rycals.^40,41 Both dantrolene and ent-verticilide inhibit RyR2 activity by prolonging the channel’s mean closed time and producing partial inhibition on RyR2 (maximal inhibition ∼50% for dantrolene⁴¹ and ∼23% for ent-verticilide). However, the dantrolene inhibition on RyR2 is dependent on calmodulin (CaM), a Ca-binding protein partner of RyR2,^41,42 and FK506-binding protein 12.6 kDa (FKBP12.6), a small RyR2 regulatory protein that binds the immunosuppressive drug FK506.^10,43 On the other hand, ent-verticilide inhibition of RyR2 is independent of CaM/FKBP12.6 binding to RyR2,¹⁵ and the activity of rycals is likewise unaffected by CaM/FKBP12.6.^44,45

It has been proposed that dantrolene binds to a peptide region between aa601 and aa620 on each of the 4 RyR2 subunits.⁴⁶ A combination of fluorescence resonance energy transfer and cryo-electron microscopy techniques has identified the dantrolene-binding sequence within the SPla and the RYanodine receptor (SPRY) domain, located adjacent to the FKBP12.6 binding site.⁴⁷ Moreover, Iyer et al⁴⁸ recently demonstrated that dantrolene interacts with additional sites on RyR1, mapping to the P1 (Ry12) domain and the lower-affinity SPRY1 domain.⁴⁸ Given that the P1 domain is highly conserved between RyR1 and RyR2 (Supplemental Material of ref⁴⁸), it is plausible that dantrolene also engages the corresponding Ry12 region in RyR2. Interestingly, the ent-verticilide binding sites were proposed to be either in the cleft of the Ry12 domain (the corner regions of the cytosolic shell) or in the interface between helical domain 1 and the SPRY3 domain—sites that lie in close proximity to the binding regions for CaM and FKBP12.6.²⁵ Notably, these regions also overlap with the reported binding sites of rycals,^44,49 and strikingly, the binding pockets for rycals and dantrolene appear nearly identical.^44,48,49 Although all 3 ligands seem to occupy overlapping regions of RyR2, it is tempting to speculate that subtle differences in ligand conformation and local interactions could account for their distinct mechanisms of action, potentially influencing isoform selectivity, and dependence on auxiliary proteins. Structural or computational studies probing these conformational landscapes may provide further insight into the molecular basis of these differences.

4.2. Clinical relevance of mode of action of ent-verticilide

Determining mode of action of ent-verticilide is crucial for identifying the disease states in which it may be therapeutically effective, enabling targeted treatment strategies. For example, in structural heart diseases, namely heart failure (HF), hyperphosphorylation of RyR2 causes Ca leak that results in reduction in the Ca transient amplitude.⁵⁰ Subsequently, SR Ca content decreased leading to impairment of cardiac contractility.⁵¹ Because a closed-channel stabilizer such as dantrolene increases SR Ca load while simultaneously decreasing SR Ca leak,⁵² it can enhance inotropy as well as prevent arrhythmia in HF. As proof-of-concept, the antiarrhythmic effects of dantrolene have been observed in animal models of HF.^9,53 Furthermore, Ca spark analysis from our previous study demonstrated that ent-verticilide reduced Ca leak and increased SR Ca content, which shared characteristics of RyR2 closed-channel stabilizers.¹⁵ As a result, it is reasonable to speculate the therapeutic potential of ent-verticilide in HF.

From a safety perspective, closed-channel stabilizers are often considered to carry proarrhythmic risks; however, we contend that this is less likely for ent-verticilide (as discussed below). Because closed-channel stabilizers elevate SR Ca content, and the likelihood of spontaneous RyR2-mediated Ca release is directly proportional to luminal SR Ca levels,⁵⁴ an increase in SR Ca content correspondingly raises the probability of spontaneous Ca release. The relationship is highly nonlinear, with the rate and amplitude of spontaneous Ca release rising sharply once the stored overload induced Ca release threshold is surpassed,⁵⁵ as luminal-triggered Ca release further activates RyR2 through positive feedback via its cytosolic activation sites.³¹ For example, tetracaine, another RyR2 closed-channel stabilizer is used clinically as a local anesthetic,⁵⁶ has been shown to increase SR content as well as increase Ca transient amplitude after each cycle of closing.^7,8 Enhanced diastolic Ca release generates a larger Na-Ca exchanger current, which elevates the probability of delayed afterdepolarizations.⁵⁷ This, in turn, increases the likelihood of triggering an action potential—a key mechanism underlying ventricular ectopic beats.⁵⁸ Unlike tetracaine, ent-verticilide increased SR Ca content while reducing SR Ca transient amplitude,¹⁵ suggesting a lower proarrhythmic potential. This further highlights the potential benefits of closed-channel stabilizers with partial RyR2 inhibition, as opposed to complete RyR2 blockers such as tetracaine. Moreover, ent-verticilide offers an additional advantage as a selective RyR2 inhibitor,¹⁵ in contrast to the pan-RyR inhibitor dantrolene,⁴⁶ which could reduce off-target effects and minimize undesirable side effects.

4.3. Carboxylic acid introduction to the ent-verticilide pentyl side chain drastically changes activity on RyR2

As part of our drug discovery efforts, we studied the SAR of novel RyR2 inhibitors derived from the ent-verticilide scaffold. All of the ent-verticilide analogs we have reported so far exhibit varying degrees of inhibitory activity on RyR2.19, 20, 21^,59 Prior SAR studies revealed the following structural features of CODs with respect to RyR2 inhibition: the backbone N-methyl amide and the ester functionality are critical for RyR2 engagement because replacement of N-methyl with N-H amide resulted in loss of activity and ester conversion to an N-H amide produced compounds with reduced potency.^19,20

Our decision to prepare activert (1) was driven by the increase in polarity provided by a carboxylic acid, and its potential to increase solubility. As a result, we introduced a carboxylic acid group to the pentyl side chain as a tactic to increase solubility. This approach is based on the premise that a single modification to the pentyl side chain is unlikely to disrupt the pharmacophore. Our previous study supports this, showing that modifying ent-verticilide’s side chain with an Alexa Fluor 568 fluorescence resonance energy transfer acceptor—despite its larger molecular weight than ent-verticilide—produced a labeled compound that retained RyR2 blocking potency.²⁵ Introducing a carboxylic acid group led unexpectedly to an analog that activated RyR2 (activert, 1) and loss of RyR2 selectivity because activert activated both RyR1 and RyR2, whereas ent-verticilide is a selective inhibitor of RyR2.¹⁵

It is possible that ent-verticilide (inhibitor) and activert (activator) share a conserved pharmacophore and binding sites, but the carboxylic acid alters its orientation, leading to distinct local conformational changes in RyR2. Our competition experiment with the Ca spark assay supported this (Supplemental Fig. 9C). For example, the carboxylic acid functionality could create an ion pair interaction with charged residues on the RyR2 structure, assuming its ionization at physiological pH, thus potentially overriding the primary binding effect. Importantly, this negatively charged moiety appears to be essential for activert’s activity, as the methyl ester of the carboxylic acid—which neutralizes its charge—abolished its ability to activate RyR2 (Supplemental Fig. 9B). Additionally, the carboxylic acid group increases the hydrophilicity of ent-verticilide’s structure and could recruit water in the vicinity of the carboxylic acid binding site, which could alter the preferred conformation for RyR2 inhibition. Another intriguing possibility is that activert binds to a site on RyR2 that is distinct from ent-verticilide’s binding site. ent-Verticilide may stabilize a RyR2 conformation that reduces activert’s binding affinity, thereby attenuating its effect. This idea is further supported by evidence that ent-verticilide may have at least 2 binding sites,²⁵ and activert can act on both RyR1 and RyR2, while ent-verticilide remains selective for RyR2, suggesting differential site accessibility and potential allosteric interactions between the 2 ligands. Although the precise mechanism of action is unclear at this point, the prospect of developing a partial activator for therapeutic use is a compelling new direction. Moreover, our findings highlight the unique versatility of the COD scaffold, which we have leveraged to discover both RyR2 inhibitors and activators. This bifurcated behavior from a single molecular scaffold stands in contrast to the rycals, which to date have yielded only RyR2 inhibitors.

4.4. The therapeutic potential for a partial RyR2 activator

As opposed to RyR2 gain-of-function mutations that caused CPVT, RyR2 loss-of-function mutations have been linked to autosomal-dominant genetic arrhythmia syndrome associated with CRDS.²² The discovery of partial activators such as activert would be valuable both as a tool compound for studying disease mechanisms and as a lead for therapeutic development. Notably, the concept of partial RyR2 activation, as exemplified by activert as potential therapeutic options for CRDS mirrors the therapeutic rationale behind partial RyR2 inhibition by agents such as ent-verticilide in CPVT. In the context of CPVT, therapeutic efficacy is achieved by attenuating RyR2 activity just enough to suppress aberrant diastolic Ca leak, without fully inhibiting RyR2, which would disrupt excitation-contraction coupling and result in fatal cardiac dysfunction. Similarly, in theory, an ideal therapeutic for CRDS would modestly enhance RyR2 activity to restore Ca release without causing excessive activation that could trigger SR Ca overload and proarrhythmic events.

The mechanism by which activert modulates RyR2 activity appears to be distinct from that of other RyR2 activators. Caffeine, a well characterized pharmacological agonist of RyR2, increases channel activity primarily by enhancing the frequency of channel openings, a mechanism that activert also appears to share.³⁶ However, caffeine induces a markedly high P_o, a property associated with its well documented proarrhythmic effects.³⁶ In contrast, activert elicits only partial RyR2 activation, resulting in a more moderate increase in P_o, which may translate to a reduced proarrhythmic risk. Another clinically relevant RyR2 modulator, amitriptyline, facilitates channel activation through stabilization of long-lasting subconductance states.²⁹ Similar to caffeine, this mechanism results in an elevated P_o and is also associated with proarrhythmic liability. Furthermore, Maxwell et al⁶⁰ recently demonstrated that remipede toxin—a bivalent peptide isolated from the venom of Xibalbanus tulumensis—is capable of activating RyR2. However, similar to caffeine, the toxin markedly increased channel P_o to supraphysiological levels, which would likely preclude its development as a viable therapeutic agent.

A limitation of activert is its poor membrane permeability (Table 1), which may limit its efficacy in vivo as the compound must cross the membrane to reach and modulate RyR2 channels. Nevertheless, activert serves as a proof-of-concept partial RyR2 activator and provides a valuable scaffold for future optimization. Together, these findings underscore partial RyR2 activation as a potentially safer and mechanistically novel approach for treating complex arrhythmogenic disorders such as CRDS. Ongoing medicinal chemistry efforts aimed at optimizing compounds such as activert could yield analogs with enhanced potency, selectivity, and membrane permeability.

5. Conclusion

In this study, we investigated the mechanism by which the unnatural form of verticilide, ent-verticilide, inhibits RyR2 activity. Using single-channel recordings, we found that ent-verticilide increased the mean closed time of RyR2 without affecting the mean open time, thus acting as a closed-channel stabilizer. We then explored structural modifications of ent-verticilide, leading to the discovery of a new analog, activert, which activated RyR1 and RyR2. Activert was confirmed to activate RyR2 through binding assays and Ca spark analysis. Single-channel recordings showed that activert decreased RyR2 mean closed time, hence destabilizing the closed-channel. Although activert exhibits poor membrane permeability that may restrict its therapeutic utility, it represents a partial RyR2 activator and a promising framework for subsequent structural refinement. Overall, these findings highlight the therapeutic potential of nonnatural enantiomers in drug discovery and emphasize the value of natural product-inspired compounds for developing innovative treatments, particularly for cardiac arrhythmias.

Conflict of interest

The authors declare no conflicts of interest.