Optical control of the cardiac rhythm with photoswitchable Na_V1.5 channel blockers

Abstract

Voltage-gated sodium channel Na_V1.5 is essential for cardiac excitability, mediating the rapid depolarization phase of the cardiac action potential (AP) and ensuring proper electrical conduction in the heart. Dysfunction of Na_V1.5 is implicated in life-threatening arrhythmias, making it a critical therapeutic target. Acting as a Na_V1.5 open-state blocker, quinidine demonstrates efficacy in arrhythmia treatment, but its low specificity restricts its clinical application. Here, we report an optopharmacological strategy that enables a precise and optical control of Na_V1.5 function by means of photoswitchable quinidine derivatives. Through systematic structural optimization, we identify azo-Q2a as a high-performance photoswitchable inhibitor, exhibiting low activity in the dark or under 480 nm light irradiation (trans isomer), while approximately 7-fold higher efficacy is observed under 365 nm light irradiation (cis isomer). Of note, azo-Q2a demonstrates exceptional selectivity for Na_V1.5 over cardiac ion channels and other Na_V1 subtypes, minimizing potential off-target effects. Furthermore, by solving the cryo-EM structure of the Na_V1.5 in complex with the cis-active isomer azo-Q2a (3.0 Å resolution), we reveal the essential binding site that is responsible for the optical control of Na_V1.5. Finally, azo-Q2a also attenuates the heart rate of living zebrafish larvae with light, showing its potential in cardiac-related research and treatment. Our work not only establishes azo-Q2a as a robust photoswitchable inhibitor for Na_V1.5 but also provides a structural blueprint for the rational design of next-generation optopharmacological antiarrhythmic agents.

This study presents azo-Q2a, a quinidine-derived photo-switchable molecule that enables optical control of Na_V1.5. It exhibits selectivity, and its binding mode is revealed by cryo-EM, offering a potential strategy for managing heart arrhythmia.

Introduction

The Na_V1.5 encoded by the SCN5A gene is the predominant subtype of the voltage-gated sodium channels family (Na_V1.1-1.9) in the heart, which plays an essential role in the excitability of cardiomyocytes¹. The channels mediate the sodium influx into cells, resulting in inward sodium current (I_Na), which induces fast depolarization of the rapid response cardiac action potential (AP). Therefore, dysfunction of Na_V1.5 channels underlies multiple cardiac channelopathies^2–4, including long QT syndrome type 3 (LQT3), Brugada syndrome (BrS), sick sinus syndrome (SSS), dilated cardiomyopathy (DCM), atrial fibrillation (AF), and progressive cardiac conduction disease (PCCD). This underscores its status as a pivotal therapeutic target for rhythm management.

Additionally, given its pivotal role in action potential generation and conduction, Na_V1.5 has long been recognized as a canonical pharmacological target for Class I antiarrhythmic drugs (Ⅰ_a). Among them, quinidine belongs to the Ⅰ_a subclass as the first antiarrhythmic drug in the world. Acting as a Na_V1.5 channel open-state blocker, quinidine inhibits peak I_Na, thereby reducing the membrane excitability of cardiomyocytes, slowing down the depolarization and conduction velocity of rapid response cardiac AP. It has been previously used for conditions such as atrial premature beats (APB), paroxysmal supraventricular tachycardia (PSVT), AF, and LQT3⁵. However, its therapeutic utility is severely constrained by extensive off-target polypharmacology. Quinidine often accompanies inhibition of the hERG channel, thereby prolonging APD, ERP, and QT interval^6,7, which increases its off-target activities⁸. Furthermore, quinidine exhibits promiscuous binding to multiple extracardiac receptors, including nicotinic acetylcholine receptors (nAChRs)⁹, 5-hydroxytryptamine receptors 3 (5-HT₃)¹⁰ and α adrenergic receptors¹¹. Its nonspecific pharmacodynamic profiles induce intolerable gastrointestinal (nausea, diarrhea) and neurological (tinnitus, cinchonism) adverse effects¹² in approximately 20–30 percents⁵ of patients during clinical application. These limitations underscore the imperative for developing targeted Na_V1.5 modulators with improved specificity.

Light has emerged as a powerful biophysical modality for spatiotemporally precise control of biological systems^13–15. Through its integration with pharmacology, optopharmacology enables non-invasive and rapidly reversible modulation of target-specific receptors and ion channels. This innovative strategy effectively circumvents the problems of off-target and dose-limiting adverse effects of traditional drugs, and thus achieves a much more precise treatment. Moreover, optopharmacology has been widely applied in ion channels^16–18, transporters¹⁹, G protein-coupled receptors^20,21, receptor-linked enzymes²². Pioneering studies have demonstrated the photopharmacological modulation of cardiac sodium channels²³, yet there remains a strong demand for novel compounds that exhibit optimal Na_V1.5 photoswitching, guided by a deeper mechanistic understanding.

In this work, we describe a photoactivated quinidine derivative with a photoswitchable azobenzene moiety incorporating, which enables reversible optical control over Na_V1.5 function and heart rate. We further characterize the binding sites of our lead compound cis-azo-Q2a using single-particle electron cryo-microscopy (cryo-EM). This provides a structural template for achieving selectivity and facilitate rational photoswitchable drug design.

Results

Rational design and synthesis of photoswitchable quinidine derivatives

As disclosed in the recently cryo-structure of Na_V1.5 in complex with quinidine²⁴, quinidine blocks the Na_V1.5 channel mainly with the quinoline moiety. While the quinuclidine group directly interrupts the central permeation pathway, the quinolone aromatic ring primarily interacts with the F1760 residue through a π-π stacking interaction. Much inspired by the above-mentioned crucial interaction, we hypothesized that quinidine derivatives with additional aryl substituents attached to the quinolone core would be beneficial for the binding affinity of Na_V1.5.

Initially, we designed three simple quinidine phenyl derivatives with phenyl substituents at the C-2 (Q1), C-3 (Q2), and C-6 (Q3) positions of the quinoline core (Fig. 1a and Supplementary Fig. 1a). Specifically, Q1 was synthesized from quinidine via a Minisci reaction²⁵ while Q2 and Q3 were prepared through Suzuki-Miyaura coupling reactions with phenylboronic acid using 3-I-QD and 6-OTf-QD as starting materials, respectively^26,27.

Fig. 1. The Na_V1.5 channel inhibition of quinidine phenyl derivatives.

a Synthesis of quinidine phenyl derivatives Q1-3. b Na_V1.5 inhibition of quinidine and its phenyl derivatives Q1-3 (10 μM).

Remarkably, whole-cell patch-clamp recordings in HEK293T cells heterologously expressing Na_V1.5 channels revealed that Q1 and Q3 exhibited significantly higher inhibition rates than quinidine at 10 µM concentration (8.8 ± 2.9% for quinidine, 69.5 ± 6.2% for Q1, and 44.7 ± 7.8% for Q3, Fig. 1b and Supplementary Fig. 1b). These results strongly suggest that enhanced π-π stacking interactions may strengthen the binding between quinidine derivatives and the Na_V1.5 ion channel.

Next, we explored photoreactive quinidine derivatives, focusing on azobenzene-based chromophores due to their widespread application. In total, nine azo-quinidine derivatives (hereafter referred to as azo-QD) were designed, featuring azobenzene moieties installed at ortho-, meta-, and para-positions with phenyl spacers at the C-2, C-3, and C-6 positions of the quinoline core (Fig. 2a).

Fig. 2. Design and synthesis of azo-QDs.

a Azologization strategy for the design of quinidine-based Na_V1.5 photoswitchable inhibitors. b Synthesis of quinidine azobenzene derivatives azo-Q1-3. c The structures and yields of azo-Q1-3.

Notably, our synthetic approach for azo-QDs utilized Suzuki-Miyaura coupling with three key intermediates: 3-I-QD, 6-OTf-QD, and 2-Br-QD (Fig. 2). The synthesis of 2-Br-QD began with the oxidation of quinidine using m-CPBA, followed by reduction with NaHSO₃, yielding N-oxide QD in 45% over two steps²⁸. Subsequent bromination via the Yu protocol afforded 2-Br-QD in 46% yield²⁹.

For 3-I-QD, direct derivatization was achieved through Suzuki-Miyaura cross-coupling with azo-BPin1 and azo-BPin2, producing the desired products in 27% and 24% yields, respectively. Meanwhile, 6-OTf-QD and 2-Br-QD were successfully coupled to furnish target compounds in 86%, 48%, 38%, and 33% yields. However, attempts to employ sterically hindered azo-BPin3 in the Suzuki-Miyaura reaction failed to yield any quinidine derivatives.

Photochemical characterization of azo-QDs

With all 6 azo-QDs in hand, photochemical characterization has been explored (Fig. 3a, b and Supplementary Fig. 1c–g). Using UV-Vis spectroscopy, we characterized the spectral properties of all cis- and trans- azo-QDs. In the dark, they existed as the thermodynamically favored trans isomer. Isomerization to the cis isomer was triggered upon UV-A irradiation (365 nm), and this process was reversed within 5 seconds upon exposure to blue light (480 nm).

Fig. 3. Photochemical characterization of azo-Q2a.

a Photochemical interconversion of trans- (480 nm illumination, blue) and cis- (365 nm illumination, violet) azo-Q2a. b The UV-Vis absorption spectra of azo-Q2a (100 μM in DMSO:H₂O = 1:100, rt) in darkness (black, trans) or at 365 nm (violet, cis) and 480 nm (blue, trans) illuminations. c A₃₄₅ and A₄₄₀ of azo-Q2a plotted over multiple cycles of alternating 365 nm (violet) and 480 nm (blue) illuminations (100 μM in DMSO:H₂O = 1:100, rt). d Conversion quantification measured by ¹H-NMR. e Thermal relaxation of azo-Q2a (rt) after converting trans-azo-Q2a to by cis-azo-Q2a illumination with 365 nm light for 5 min.

Taken azo-Q2a as an example, we observed a rapid and reversable photoswitch upon oscillating between 365 and 480 nm illumination over cycles (Fig. 3c). Using ¹H-NMR and ¹³C-NMR spectroscopy, we assessed the relative concentrations of the two isomers in equilibrium after 365 and 480 nm illumination (Fig. 3d, Supplementary Figs. 2, 3). After illumination at 365 nm, trans isomer azo-Q2a was efficiently isomerized to cis isomer (86%); 480 nm illumination resulted in efficient isomerization back to the trans isomer (79%). While we noted that the isosbestic point shifts slightly upon illumination at 480 nm, which is small relative to the main spectral features, this might result from minor side- or photoproduct formation or from changes in solvent polarity and ionic strength.

Furthermore, the photoisomerization rates of azo-Q2a were measured. Under 365 nm UV light at 8500 lux, the half-life was 28 min, whereas under 480 nm blue light at 1500 lux, the half-life was 3.5 min. The quantum yields of azo-Q2a were determined to be 0.075 under 365 nm irradiation and 0.32 under 480 nm irradiation. These results demonstrate that azo-Q2a exhibits efficient and controllable photoreactive behavior across different irradiation wavelengths³⁰ (Supplementary Fig. 4 and Supplementary Table 1).

Finally, we examined the chemical stability of azo-Q2a. Under in vitro conditions, azo-Q2a remained remarkably stable, exhibiting a half-life (t_1/2) of 107 h in the absence of reducing agents (Fig. 3e). In contrast, exposure to 1 mM glutathione (GSH) triggered rapid cis-to-trans isomerization with a half-life of 59 min, driven by the reversible reduction of the -N = N- bond. At a higher GSH concentration (10 mM), azo-Q2a underwent predominant reductive degradation with a half-life of 143 min, indicating a shift in the reaction pathway under strongly reducing conditions (Supplementary Fig. 5).

Optical control of Na_V1.5 channels with photoswitchable azo-QDs

To determine whether azo-QDs support optical control of Na_V1.5 channels, we initially profiled the concentration-dependent inhibition activities of quinidine and six derivatives to Na_V1.5 current in the dark or under 365 nm illumination, using whole-cell patch-clamping recordings in HEK293T cells heterologously expressing Na_V1.5 channels. (Fig. 4a–d and Supplementary Fig. 6a–e).

Fig. 4. azo-Q2a reversibly inhibits Na_V1.5 channels with light in HEK293T cells.

a The Na_V1.5 inhibition (IC₅₀) of quinidine azobenzene derivatives azo-Q1-3, either in darkness (gray, trans) or at 365 nm (violet, cis) illumination, IC₅₀ value and its 95% confidence interval obtained by specific fitting the Hill equation with the dose-dependence values measured in (c, d) and Supplementary Fig. 6a-e. Quinidine (n = 8, 8), azo-Q1a(n = 6, 7), azo-Q1b (n = 6, 6), azo-Q2a (n = 9, 12), azo-Q2b (n = 6, 6), azo-Q3a (n = 8, 5), azo-Q3b (n = 7, 6). Each n represents a patch of independent cells. Error bars represent 95% Confidence Interval (CI). b–h Whole-cell patch clamp recordings of Na_V1.5 currents through the application of azo-Q2a and quinidine in darkness (black) or at 480 nm (blue) and 365 nm (violet) illumination. b The schematic of experiment design (left, created in BioRender. Peng, C. (2026) https://BioRender.com/0iomjnp) and representative current traces of Na_V1.5 blocked by varied concentrations of azo-Q2a and quinidine (right). 0 μM, 10 μM, and 100 μM in black, azure, and orange, respectively. c, d concentration-dependent response curve of azo-Q2a (c; n = 9,12) and quinidine (d; n = 8, 8) in the dark or at 365 nm light. e Representative current traces and a diagram of recording protocol of Na_V1.5 activation (left). I–V (e) and g–V (f) relationship in 10 µM azo-Q2a under only 365 nm (n = 15) or 480 nm (n = 12) illumination and without azo-Q2a (control, n = 49). Statistical evaluation by ordinary one-way ANOVA. e Control vs. 480 nm (P = 0.92), Control vs. 365 nm (*P = 0.026). f Control vs. 480 nm (P = 0.41), Control vs. 365 nm (P = 0.92), 480 nm vs. 365 nm (P = 0.78). *P < 0.05, ns, not significant (P > 0.05). g Representative current traces of reversible photoswitch of Na_V1.5 by 10 µM azo-Q2a. Selected are the last triggered current responses under 365 nm and 480 nm light cycles, each following the protocol showed on b at 1 Hz for 15 times. Parallel lines represent time gaps between steps. h Peak Na_V1.5 currents of g over multiple cycles. Each n represents a patch of independent cell. Error bars represent mean ± SEM.

When holding the cellular membrane potential at −120 mV, the expected Na⁺ currents were observed at the typical voltage steps³¹ (−120 to −20 mV). All of the examined azo-QDs in the cis isomer conferred stronger inhibition activity than quinidine, and azo-Q1a, azo-Q2a, and azo-Q3b displayed a significant increase in their IC₅₀ values after 365 nm illumination. (Fig. 4a–c and Supplementary Fig. 6a–e). Among the examined azo-QDs, azo-Q2a showed the most prominent alteration (IC₅₀ = 11.76 µM in cis; 81.58 µM in trans, Fig. 4c), while quinidine had no obvious difference (IC₅₀ = 35.87 µM under 365 nm; 41.21 µM in dark, Fig. 4d). IC₅₀ values are detailed in Supplementary Table 2.

We measured current-voltage (I-V) steps between −120 and +40 mV (5 mV steps) in the presence of 10 µM azo-Q2a (Fig. 4e). In its cis isomer, azo-Q2a strongly suppressed inward currents conducted by Na_V1.5 channels, whereas the trans isomer had no discernible effect. In addition, neither isomer altered the activation property of Na_V1.5, assessed as conductance-voltage (g-V) relationships and the V₅₀ value (Fig. 4f, Supplementary Table 3). Note that blockade of Na_V1.5 channels was readily reversible as we repeatedly photoswitched azo-Q2a between its trans and cis isomer (Fig. 4g, h).

azo-Q2a is highly selective and significantly reduces the hERG toxicity of quinidine

To further evaluate the selectivity profile of azo-Q2a among other sodium channel subtypes (Na_V1.2–Na_V1.6) highly expressed in the central nervous system and muscle tissues (Supplementary Fig. 7a). At a concentration of 10 μM, cis-azo-Q2a exhibited potent inhibition of Na_V1.5 (61.1 ± 4.8%, use-dependent, Supplementary Fig. 6f) but minimal activity (<10%) against all other subtypes tested. In contrast, trans-azo-Q2a under dark conditions showed negligible suppression of Na_V1.5 (2.5 ± 1.6%) yet significantly inhibited Na_V1.3 (36.2 ± 6.1%). Notably, this off-target effect on Na_V1.3 was effectively reversed upon optical illumination, which switches the compound to cis-configuration.

These results underscore the light-dependent selectivity of azo-Q2a for Na_V1.5. Given that Na_V1.5 serves as the predominant sodium channel in cardiac tissue, spatially confined illumination enables the high-affinity cis-azo-Q2a to selectively modulate heart function while minimizing potential off-target effects in neural or other systems.

In addition, recalling that quinidine’s off-target interactions that inhibit hERG channels lead to serious cardiotoxicity⁷, we investigated the potential inhibition activity of azo-Q2a towards other ion channels known to contribute to cardiac action potential, including K_V4.3 (I_to), K_ir2.1 (I_K1), hERG (I_Kr) and Ca_V1.2 (I_CaL) channels (Fig. 5a). The experiments monitored the currents of these channels under both 365 and 480 nm illumination. Currents of these ion channels were recorded using whole-cell patch-clamping in HEK293T and CHO cell lines expressing cardiac ion channels (see Methods section for details).

Fig. 5. azo-Q2a selectively optical controls Na_V1.5 channels in cardiomyocytes.

a, b Whole-cell patch clamp recordings of Na_V1.5, K_V4.3, K_ir2.1, hERG, and Ca_V1.2 currents through application of 10 µM azo-Q2a and quinidine at 480 nm (blue), 365 nm (violet) illumination with no inhibition (control, black). a Representative current traces and diagrams of recording protocols. b Normalized inhibition rate. Na_V1.5 (n = 9, 11, 7, 7), K_V4.3 (n = 8, 7, 7, 7), K_ir2.1 (n = 11, 11, 8, 7), hERG (n = 15, 16, 8, 7) and Ca_V1.2 (n = 8, 6, 6, 6). P value < 0.0001; P value = 0.89, 0.95, 0.68, 0.09, 0.76, 0.74, 0.69, 0.19, 0.48 (from left to right colunm). c Schematic of primary cardiomyocytes patch experiment. Created in BioRender. Peng, C. (2026) https://BioRender.com/dg6caub. d Representative current traces of reversible photoswitch of Na⁺ current in primary ventricular myocytes induced by 10 µM azo-Q2a. Selected are the last triggered current responses under 365 nm and 480 nm light cycles, each following the protocol showed on a (for Na_V1.5) at 1 Hz for 15 times. e, Peak Na⁺ currents of d over multiple cycles. d–f Recordings of Na⁺ and K⁺ currents with the same treatment of a, b. f Representative current traces and diagrams of recording protocols. g Normalized inhibition rate. n = 5, 7, 5, 6, 10, 9, 5, 5 (from left to right column). P value = 0.0006, 0.66, 0.65, 0.87 (from left to right column). h Quantification of photoswitch induced by 10 µM azo-Q2a. n = 6, 5, 9, 4, 7, 5, 5 (from left to right column). Na_V1.5 vs. Na⁺ current (P = 0.34), the rest group P < 0.0001. Statistical evaluation by ordinary one-way ANOVA. Data are statistically compared with Na_V1.5 group. a–g, Each n represents a patch of independent cells. Each data point was from biological sample repeats (n ≥ 3). Statistical evaluation by a two-tailed unpaired t test. ***P < 0.001, ns, not significant (P > 0.05). Error bars represent mean ± SEM.

The inhibition rate of 10 µM azo-Q2a for Na_V1.5 channels was significantly different between the cis (61.1 ± 4.8%) and trans (2.5 ± 1.6%) isomers, whereas both azo-Q2a isomers exerted in same inhibition activity against K_V4.3, K_ir2.1, hERG and Ca_V1.2 channels (Fig. 5b). Quantification of the inhibition showed that azo-Q2a endows light sensitivity only to Na_V1.5 but has negligible effects on K_V and Ca_V channels, yielding photoswitching efficacies of 33.1 ± 4.3% for Na_V1.5 and far less for the rest channels, of 5.2 ± 1.7% (K_V4.3), 3.6 ± 4.2% (K_ir2.1), 7.9 ± 4.5% (hERG), 7.7 ± 5.4% (Ca_V1.2) (Fig. 5h and Supplementary Table 4). To compare the difference in hERG channel inhibition activity between azo-Q2a and quinidine, we repeated these assessments through application of 10 and 30 µM quinidine. As expected, quinidine showed comparable pharmacological behavior across all tested ion channels without optical control property (Fig. 5b and Supplementary Fig. 7b–d). Notably, at 10 µM, cis-azo-Q2a inhibited Na_V1.5 and hERG currents by 61.1% and 45.5%, respectively, yielding a Na_V1.5/hERG inhibition ratio of 1.34, whereas quinidine produced 8.8% and 71.7% inhibition (ratio 0.12). This corresponds to an approximately 11-fold increase in the Na_V1.5/hERG inhibition ratio, indicating an overall enhancement of functional selectivity relative to quinidine (Fig. 5b and Supplementary Table 5). In addition, we further evaluated the concentration-response relationship of azo-Q2a and quinidine on hERG channels under 365 and 480 nm illumination, which showed that the IC₅₀ value of azo-Q2a on hERG was 16-fold lower than that of quinidine (Supplementary Fig. 7e, f).

Together, these results demonstrate that azo-Q2a possesses high selectivity for Na_V1.5 channels among cardiac ion channels and other Na_V1 subtypes, significantly reducing the hERG channel inhibition activity observed with quinidine. These characteristics make azo-Q2a a promising candidate for the development of photoactivated antiarrhythmic therapies.

azo-Q2a reversibly inhibits Na⁺ currents in rat primary cardiomyocytes

We next investigated azo-Q2a as an agent for the selective optical control of Na_V1.5 channels in rat primary cardiomyocytes. We recorded voltage-gated Na⁺ and K⁺ currents from primary ventricular myocytes under 365 and 480 nm illumination (Fig. 5c).

Consistent with our observations in the transient transfection cell experiments, Na⁺ current in cardiomyocytes was reduced at 365 nm illumination after the addition of azo-Q2a (10 µM), and this was reversed by blue light (480 nm) (Fig. 5d). The selective optical control of Na_V1.5 channels was over multiple on/off cycles (Fig. 5d, e); the photoswitching efficacy was 24.3 ± 3.3%, showing no significant difference from the 33.1 ± 4.3% compared to HEK293T cells expressing Na_V1.5 channels (Fig. 5h). No significant inhibition on K⁺ currents were observed under the same experimental conditions (3.0 ± 2.4%), supporting the selectivity of azo-Q2a towards Na_V1.5 channels (Fig. 5f–h). Additionally, quinidine (in 10 and 30 µM) was applied as a control, which showed insignificant optical control of both Na⁺ (0.8 ± 0.8%) and K⁺ currents (0.9 ± 1.9%, in 10 µM), (Supplementary Fig. 7b–d). To further exclude potential effects of UV-induced ROS, we performed illumination-only controls (Supplementary Fig. 8a), which showed no changes in sodium currents, consistent with the unchanged inhibition of quinidine under both wavelengths.

Owing to the significant role of Na_V1.5 in the action potential of myocardial cells, we directly evaluate the effect of azo-Q2a on cardiac electrophysiology performing whole-cell current-clamp recordings on primary cardiomyocytes to assess the impact of trans and cis-azo-Q2a on cardiac action potential (Supplementary Fig. 8b–e). The results demonstrated that cis-azo-Q2a significantly reduced the dV/dt_max by 47.2 ± 8.7%, whereas trans-azo-Q2a showed only a modest inhibition of 16.6 ± 1.1%. Furthermore, we observed that cis-azo-Q2a shortened the action potential duration at 50% repolarization (APD₅₀) by 55.2 ± 4.9% whereas trans-azo-Q2a produced only a modest inhibition of 16.6 ± 3.4%. We attribute this shortening primarily to the inhibition of the late sodium current (I_{Na, L}), which is known to modulate repolarization.

Cryo-EM Structure basis for azo-Q2a optical control of Na_V1.5 channels

To assess azo-Q2a binding, we obtained a cryo-EM structure for Na_V1.5 in complex with azo-Q2a at the resolution of 3.0 Å (Na_V1.5S-azo-Q2a) (Supplementary Table 6). Illuminated by 365 nm light, azo-Q2a was photo-induced as the cis isomer before sample preparation. Note also that, seeking to optimize protein expression, we truncated the carboxyl-terminal domain and linker between repeats Ⅰ-Ⅱ and Ⅱ-Ⅲ of Na_V1.5³² (referred to as Na_V1.5S, Supplementary Fig. 9).

We first assessed the inhibition activity of quinidine on Na_V1.5S using whole-cell patch-clamping recording, yielding an IC₅₀ value of 44.66 μM in the dark and of 43.80 μM under 365 nm illumination; these values do not differ from the values for Na_V1.5^WT (Supplementary Fig. 10a, c and Supplementary Table 7). We then examined the optical control of azo-Q2a towards Na_V1.5S following the same recording protocols and conditions used for Na_V1.5^WT and found that azo-Q2a showed the same inhibition activity and optical control property for Na_V1.5S and Na_V1.5^WT (IC₅₀ = 10.44 µM for cis; 75.41 µM for trans, Supplementary Fig. 10b, d), suggesting that truncation of Na_V1.5 does not interfere with the interaction of quinidine or azo-Q2a in the pore domain.

The unambiguous EM densities for cis-azo-Q2a are shown in the blue mesh support that the cis-azo-Q2a molecule bind to the pore domain (Fig. 6a, Supplementary Fig. 11a). Similar to quinidine, cis-azo-Q2a is coordinated by both polar and hydrophobic residues from repeats Ⅰ, Ⅲ, and Ⅳ, but distanced from repeat Ⅱ (Fig. 6a–c). In addition, upon comparing our complex structure of Na_V1.5S with the reported Na_V1.5-quinidine structure³¹ (PDB ID: 6LQA), we found that the binding sites of azo-Q2a and quinidine differ in the central cavity of the pore domain (Fig. 6b–d): quinidine is likely hydrogen-bonded to Gln371, Thr1417, and Ser1759 and appears to interact with Phe1760 through π-π stacking; cis-azo-Q2a is surrounded by multiple hydrophobic residues from the S6 segments (Fig. 6b, c), among which Val405 appears to interact with the quinolone aromatic ring through a hydrophobic interaction (Fig. 5b, c), while Phe1760 plays an essential role connecting with the azobenzene group through π-π stacking (Fig. 6b, c). The interactions between Phe1760 and the quinoline moiety of quinidine changed to the azobenzene portion of cis-azo-Q2a, while the S6_Ⅳ of Na_V1.5S undergoes an α to π helix transition (Fig. 6e, Supplementary Fig. 11b).

Fig. 6. Molecular basis for azo-Q2a coordination.

a Cis-azo-Q2a is positioned beneath the ion selectivity filter. Shown here are two perpendicular views Na_V1.5S-azo-Q2a. The four repeats of Na_V1.5S are colored grey, green, yellow and light blue, respectively. The Ⅲ-Ⅳ linker is colored orange, azo-Q2a is colored brown, and the sodium ion is colored purple. b Coordination of azo-Q2a by residues from repeats Ⅰ, Ⅲ, and Ⅳ. Potential hydrogen bonds are indicated by red dashed lines. c A schematic representation of the quinidine-azo binding pocket. Residues within 4 Å of quinidine-azo are shown. d Structural comparison of Na_V1.5S-azo-Q2a and Na_V1.5-quinidine. The structures of azo-Q2a-bound Na_V1.5S (domain-colored) and quinidine-bound Na_V1.5 (silver, pdb code: 6LQA) are nearly identical. The root-mean-square deviation between them is 0.716 Å over 975 Cα atoms. The binding pose of azo-Q2a and quinidine is different in the central cavity. e Structural differences between Na_V1.5S-azo-Q2a and Na_V1.5-quinidine. Besides the binding pose of azo-Q2a and quinidine, the S6_Ⅳ of Na_V1.5S undergoes an α to π helix transition. f The schematic of representative current traces of Na_V1.5 mutations V405A (left) and F1760A (right) blocked by varied concentrations of azo-Q2a. 0 μM, 10 μM, and 100 μM in black, azure, and orange, respectively. g, h Concentration-dependent response curve of V405A (d; n = 7, 6), F1760A (h; n = 6, 6) by azo-Q2a in the dark (black) or at 365 nm (violet) light. Each n represents a patch of independent cells. Error bars represent mean ± SEM.

We generated Na_V1.5 variants bearing V405A and F1760A mutations, expressed them in HEK293T cells, and assayed azo-Q2a inhibition activity in the dark or under 365 nm illumination. We assessed the inhibitory effect of 100 μM quinidine on the V405A and F1760A mutants. The results demonstrated that the functional impairment was more pronounced in the F1760A mutant compared to the WT, with inhibition rates of 77.6 ± 3.2 % (WT), 51.2 ± 2.3% (V405A), and 32.6 ± 1.8% (F1760A), (Supplementary Fig. 7g). This finding is consistent with previous cryo-EM structural analyses, which revealed that the quinoline ring of quinidine forms critical interactions with the F1760 residue³¹.

Assessing the calculated concentration-responses supporting our Na_V1.5S-azo-Q2a structural insights: Both mutations V405A and F1760A resulted in a significant rightward shift of the IC₅₀ curve for cis-azo-Q2a (IC₅₀ > 100 μM, Fig. 6f–h and Supplementary Table 8), indicating a substantial reduction in its inhibitory potency. Collectively, these findings confirm that Val405 and Phe1760 are critical residues for the binding and inhibition of Na_V1.5 by cis-azo-Q2a, thereby establishing them as essential molecular determinants for the light-dependent control mediated by azo-Q2a.

azo-Q2a reversibly attenuates heart rate and terminates ISO-induced tachycardia in vivo in zebrafish larvae

To assess the capability of azo-Q2a to confer optical heart rate control in vivo, we used zebrafish larvae (2 days post-fertilization) as the animal model (Fig. 7a), for which light scattering is known to be low³³. The Na⁺ current (I_Na) generated by Na_V1.5 channels is essential for maintaining a normal heart rate. Heart rate was monitored by counting the number of sequential contractions in 10-30 s intervals under a dissecting microscope.

Fig. 7. azo-Q2a optical regulates heart rate in zebrafish.

a The schematic of recording heart rate. Created in BioRender. Peng, C. (2026) https://BioRender.com/7pejkc8. b The heart rhythm of zebrafish larvae in dark (gray), after 365 nm illumination for 5 min (violet) or 365 nm following 480 nm illumination for both 5 min (blue) with no application (control, n = 22, 18, 16), 10 µM quinidine (n = 22, 22, 21) or 10 µM azo-Q2a (n = 25, 11, 11). All data are normalized to mean of the control experimental group recording in dark (160 bpm). Each data point was from biological sample repeats (n ≥ 3). Statistical evaluation by ordinary one-way ANOVA. P value: >0.99, =0.98, =0.99, =0.97, =0.99, =0.99, <0.0001, = 0.97, <0.0001. (from left to right colunm). ***P < 0.001, ns, not significant (P > 0.05) Error bars represent mean ± SEM.

The larvae were exposed to different treatments via soaking (Control; 10 μM azo-Q2a and 10 μM quinidine) for 1-1.5 h in the dark³⁴, and then heart rate was recorded 30 min after 365 or 480 nm light exposure using video microscopy (Fig. 7a). We illuminated control zebrafish larvae that we kept under dissimilar experimental conditions for 5 min (at 365 nm or at 365 nm followed by 480 nm light both continuing for 5 min), aiming to estimate optical effects on the heart rate (Fig. 7a).

We first examined the heart rate of larvae treated with no inhibitor as a control: these showed no significant change after different light exposure, suggesting that the light itself does not affect the heart rate (Fig. 7b). The average heart rate of control experiments measured in the dark was used as baseline to normalize each detection value. After perfusion of trans-azo-Q2a (10 μM) lasting for 1-1.5 h in the dark, there was no difference in the calculated heart rate (98.8 ± 1.4%) compared to the control (Supplementary Movie 1) treated with DMSO. However, we did observe a significant slowing of the heart rate for these larvae after 365 nm illumination (56.5 ± 4.3%, Supplementary Movie 2), and this slowing was reversed by the subsequent 480 nm illumination (103.2 ± 2.0%, Fig. 7b, Supplementary Table 9 and Supplementary Movie 3).

We also examined larvae perfused with quinidine in the dark (90.8 ± 1.4%), under 365 nm illumination (93.9 ± 0.9%), and after subsequent 480 nm illumination (91.6 ± 1.3%), (Fig. 7b and Supplementary Table 9). These results are consistent with previous reports for quinidine’s inhibition activity, demonstrating that azo-Q2a confers optical control and has higher inhibition of the heart rate in vivo compared with quinidine³⁵.

To further evaluate the therapeutic potential of azo-Q2a against arrhythmias, we established a tachycardia model in zebrafish larvae (3–4 dpf, untreated control shown in Supplementary Movie 4) by incubating them in darkness with 10 μM isoproterenol (ISO) for 100 min, which significantly increased heart rate³⁶ (Supplementary Movie 5). Using the same treatment and illumination protocol as described above, we found that both 10 μM cis-azo-Q2a (35.3 ± 2.6%, Supplementary Movie 6) and 100 μM quinidine (Supplementary Movie 7) markedly attenuate heart rate in ISO-pretreated larvae, whereas 10 μM trans-azo-Q2a (Supplementary Movie 8) produced only a mild reduction in heart rate (4.4 ± 1.2%, Supplementary Fig. 12). These results demonstrate that light-induced isomerization of azo-Q2a can effectively terminate ISO-induced tachycardia in vivo, thereby supporting its potential relevance for Na_V1.5-related arrhythmias.

Overall, these results indicate that the optical control phenomena of electrophysiological experiments obtained in vitro can be reproduced in living animals and, moreover, that azo-Q2a can reversibly control heart rate.

Discussion

As the first antiarrhythmic drug in the world, quinidine has been used in the treatment of almost all cardiac arrhythmias since the early twentieth century, but in the last two decades, it decreases in clinical prescriptions due to its lack of specificity, which leads to intolerable adverse effects in the gastrointestinal tract and nervous system. To date, few teams have chosen to improve the drug properties of quinidine through structural modification. Instead, we here report on an optopharmacology strategy that not only enables specific optical control of Na_V1.5 channels, but also enhances its activity and selectivity. The best-performing photoswitchable quinidine derivative modified by azobenzene (azo-Q2a) showed 3-fold higher efficacy and only demonstrated optical control of Na_V1.5 function among all our evaluated cardiac ion channels. The high selectivity of it combined with the high spatiotemporal specificity of optopharmacology, gives azo-Q2a a predictable potential to reduce the adverse effects of quinidine in clinical treatment. We also performed the cryo-EM structure of Na_V1.5S in complex with cis-azo-Q2a, revealing the binding sites through which azo-Q2a achieves high inhibition efficiency and optical control towards Na_V1.5S and providing the chance for further precise modification of azobenzene.

To gain insight into the potential optical control mechanism of azo-Q2a, we docked trans-azo-Q2a into the Na_V1.5 structure. In the absence of spatial constraints, the predicted binding poses of the trans isomer were displaced from the well-resolved cis-azo-Q2a binding site (Supplementary Fig. 11b). When docking was performed with the pose constrained to match that of cis-azo-Q2a, the principal divergence between the two binding modes was localized to the azobenzene moiety (Supplementary Fig. 11c, molecular docking data for cis-azo-Q2a and trans-azo-Q2a are provided in Supplementary Data 1 and 2, respectively). In addition, the cryo-EM map shows that the local density for the azobenzene group in the docked trans isomer is poor and that it lacks stable interactions with the surrounding residues (Supplementary Fig. 11d). Such structural plasticity may account for the different binding affinities between cis and trans isomers of azo-Q2a for Na_V1.5.

Light is unsurpassed in its ability to control biological systems with high spatial and temporal resolution. Many photoswitchable molecules have been reported in recent years, such as the photostatins, which act as inhibitors of microtubule dynamics³⁷. Our work demonstrates the feasibility of using optopharmacology to control the activity of natural products and to repurpose existing drugs. This approach enhances the pharmacological properties of a drug while conferring additional functions.

Notably, we verified the capability of azo-Q2a to confer optical control of heart rate in vivo. On 2 dpf living zebrafish larvae, azo-Q2a enabled optical control of the heart rate, demonstrating its potential in the treatment of NaV1.5-related arrhythmias. It should be noted that zebrafish larvae are highly sensitive to repeated light exposure. Consequently, extended recovery periods (30 minutes) must be incorporated between light cycles to minimize disturbance³⁴. This requirement poses an inherent limitation on the repeatability of optical manipulations for controlling heart rhythm in this model. Nevertheless, it is incontestable that using the basic azobenzene motif means the optical wavelength of photoswitchable flipping is mostly located near 360 ~ 400 nm and 480 ~ 550 nm³⁸. To achieve noninvasive optical control of heart rate in other model animals like mice or even in human require it to have a red shift to nearly 600 nm³⁹. Though modification of azobenzene by the ortho halogen or methoxyl substitution can be available to carry out the expectant red shift, our improved azo-Q2a of which the maximum absorbance is in the visible light region, failed to maintain the remarkable optical control of Na_V1.5 channel (Supplementary Fig. 13). Only the cis-to-trans isomerization wavelength shifted from blue (480 nm) to green (550 nm) light, while the trans-to-cis photoactivation wavelength remained in the UV range. While electrophysiological tests revealed that the inhibitory efficacy of cis-azo-4F (8.3 ± 4.6% in 10 μM) on Na_V1.5 currents was unfortunately reduced compared to cis-azo-Q2a (61.1 ± 4.8% in 10 μM). Therefore, further modifications of the azobenzene motif or other photoswitchable moiety is required.

In conclusion, the quinidine-azobenzene (azo-Q2a) is a promising photoswitchable molecule targeting Na_V1.5 channel, which is worthwhile to future study on cardiac diseases that are especially caused by Na_V1.5 mutation.

Methods

Ethics statement

All animal procedures were conducted in strict accordance with the “Guide for the Care and Use of Laboratory Animals” and the “Principles for the Utilization and Care of Vertebrate Animals”. All experimental protocols were approved by the Institutional Animal Care and Use Committee of Peking University Health Science Center (DLASBD0572). All zebrafish (AB strain) used in this study were raised and handled according to protocol (IMM-XiongwW-3) approved by the Institutional Animal Care and Use Committee at Peking University, which is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC). Every effort was made to minimize animal suffering and to reduce the number of animals used.

Chemical synthesis and photochemical characterizations

azo-QDs were synthesized and characterized by standard chemical methods. The UV-visible absorption spectra of the cis and trans isomers were determined by UV-Vis spectroscopy. 365 nm and 480 nm LED light were used to perform trans ↔ cis isomerization in vitro. Ratio of the trans and cis isomerization and half-life of cis isomerization were determined by NMR spectra. Full design, synthesis, and photochemical characterization of the azo-QDs is detailed in the Supplementary Information.

Cell culture and transfection

The human embryonic kidney cells (HEK293 and HEK-293T) and Chinese hamster ovary cells (CHO) were obtained from ATCC. HEK293 and HEK-293T cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM, Gibco) supplemented with 15 % (volume/volume) Fetal Bovine Serum (FBS, PAN-Biotech) at 37 °C and 5 % CO₂. In addition, due to poor adherence of HEK293T stable cell lines expressing K_V4.3 channels and K_ir2.1 channels, 0.1 mg/mL Poly-D-lysine (Sigma) was perfused in culture dishes prior to cell culture, and the ratio of FBS increased to 20 %. CHO cells were maintained in Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (1:1, DMEM/F12, SinoDetech) supplemented with 10 % (volume/volume) FBS.

Cells with 70 ~ 80 % confluence were transiently transfected with plasmid DNA using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Then the cells were cultured for 20 ~ 48 h and subjected to experiments.

Voltage-clamp recordings

Voltage-clamp recordings were obtained using a HEKA EPC-10 patch-clamping amplifier (HEKA Electronic), and the data were acquired by the PatchMaster program (HEKA Electronic). For whole-cell patch-clamping recording, patch pipettes were pulled from thin-walled borosilicate glass (Sutter Instrument), polished, and had resistances of 1.5 to 2.5 MΩ in the experimental solutions. In addition, all recordings were accepted only when series resistance was <6 MΩ and compensated by 70%, and seal resistance exceededå 1 GΩ. To evaluate the inhibition effect of quinidine and azo-QDs, compounds were delivered by gravity perfusion. The illumination during whole-cell patch-clamping recording was provided by LumiCite 9100 LED (OPLENIC). For a 480 nm wavelength, 100% light intensity was used (45000 lux), and for a 365 nm wavelength, 35% light intensity was used (1200 lux).

As for the voltage-clamp recording analyses, all data were reported as mean ± SEM. Data analyses were performed using Origin 2022 (Origin Lab), Excel 2016 (Microsoft), and GraphPad Prism 10 (GraphPad Software). Inhibition curves were generated using a Hill Eq. (1)

$$\frac{I}{I_{\max }}=\frac{1}{1+{10}^{\left(\log {\mathrm{IC}}_{50}-\left[\mathrm{C}\right]\right)\times H}}$$ 1

where I is the current at different compounds concentrations, I_max is the maximal current of ion channels without inhibitor application, [C] is the logarithmic concentration, IC₅₀ is the half-maximal inhibition concentration and H is the Hill coefficient.

Sodium channel

For sodium currents recording, the pipette solution contained 140 mM CsF, 10 mM HEPES, 1 mM EGTA, 10 mM NaCl (pH = 7.3 with CsOH and an osmolarity of ~295 mOsm/L). The bath solution contained 140 mM NaCl, 3 mM KCl, 10 mM HEPES, 10 mM Glucose, 1 mM CaCl₂, 1 mM MgCl₂ (pH = 7.3 with NaOH and an osmolarity of ~310 mOsm/L). In whole-cell recordings of Na_V1.5 currents, the cells were held at −120 mV and the inward peak sodium currents were elicited by a 50 ms step to −20 mV. Recordings from other sodium channel isoforms (Na_V1.2–Na_V1.6) were performed under the same holding conditions (−120 mV) but using subtype-specific test pulses: 0 mV for Na_V1.2 and Na_V1.3 (50 ms), −30 mV for Na_V1.4 (20 ms), and −10 mV for Na_V1.6 (50 ms). In addition, the current-voltage (I–V) relationships were obtained by a 50 ms step from −120 to +40 mV in 5 mV increments. To calculate the voltage dependence of activation, conductance G was fitted with the Boltzmann Eq. (2)

$$\frac{G}{G_{\max }}=\frac{1}{1+{\mathrm{e}}^{\left(V_m-V_{\frac{1}{2}}\right)}/-k}$$ 2

where G is conductance and G_max is the maximum conductance between–120 and+40 mV. V_m is the stimulus potential. V_1/2 indicates the voltage at half-maximal activation, and k is a slope factor describing the voltage sensitivity of the channel. k is the slope factor. G are calculated from the I–V relationships according to Eq. (3)

$$G=\frac{I}{V_{\mathrm{m}}+E_{\mathrm{Na}}}$$ 3

where I is the peak current, G is conductance, V_m is the stimulus potential, E_Na is the equilibrium potential, which is the calculated reversal potential based on our whole-cell recording of Na_V1.5 I-V response curve. Significance of fitted V_1/2 compared to control was analyzed using the extra sum-of-squares F test.

Potassium channel

For potassium currents recording, the pipette solution contained 130 mM KCl, 10 mM HEPES, 5 mM EGTA, 1 mM MgCl₂, 5 mM Mg-ATP (pH = 7.3 with KOH and an osmolarity of ~295 mOsm/L). The bath solution contained 138 mM NaCl, 4 mM KCl, 10 mM HEPES, 10 mM Glucose, 2 mM CaCl₂, 1 mM MgCl₂, 0.33 mM NaH₂PO₄·2H₂O (pH = 7.3 with NaOH and an osmolarity of ~310 mOsm/L). In whole-cell recordings of K_V4.3 currents, the cells were held at −80 mV then currents were elicited by a 600 ms step to +60 mV. For hERG currents, the cells were held at −80 mV then currents were elicited by a step to +20 mV following a step to −40 mV, each stimulation lasting for 2 s^40,41. For K_ir2.1 currents, the cells were held at −80 mV then currents were elicited by a 100 ms step to −120 mV.

Calcium channel

For calcium currents recording, the pipette solution contained 135 mM K-gluconate, 10 mM HEPES, 5 mM EGTA, 2 mM MgCl₂, 5 mM NaCl, 4 mM Mg-ATP (pH = 7.3 with KOH and an osmolarity of ~295 mOsm/L). The bath solution contained 105 mM NaCl, 30 mM TEA-Cl, 10 mM BaCl₂·2H₂O, 10 mM HEPES, 10 mM Glucose, 5 mM CsCl, 4 mM KCl, 1 mM MgCl₂ (pH = 7.3 with NaOH and an osmolarity of ~310 mOsm/L). In whole-cell recordings of Ca_V1.2 currents, the cells were held at −80 mV then currents were elicited by a 200 ms step to −10 mV.

Isolate and culture rat ventricular primary cardiomyocytes

Ventricular primary cardiomyocytes were isolated from 1- to 2-day-old Sprague-Dawley rats⁴² provided by Peking University Health Science Center Department of Laboratory Animal Science. All animals were handled in strict accordance with the “Guide for the Care and Use of Laboratory Animals” and the “Principles for the Utilization and Care of Vertebrate Animals,” and all animal work was approved by the Institutional Animal Care and Use Committee of Peking University Health Science Center (DLASBD0572). Each effort was made to minimize animal suffering and the number of animals used. The experiments were blind to viral or drug treatment conditions during behavioral testing. The gender of the fetal rats was not screened during the experiment. Briefly, primary cardiomyocytes were isolated with 0.05% trypsin (Gibco) and 0.1% collagenase II (Sigma). Cardiomyocytes were separated from the fibroblasts by pre-plating the digested cell suspension for 2 h. Cells were maintained in DMEM (Gibco) supplemented with 10% (volume/volume) FBS (PAN-Biotech) and antibiotics (100 U/ml penicillin and 100 mg/ml streptomycin) (Gibco) at 37 °C and 5% CO₂.

The voltage-gated whole-cell patch-clamping recording of sodium and potassium currents in primary cardiomyocytes follows the protocol in HEK293T cells. For cardiac action potential recordings, data were collected using a Multiclamp 700B amplifier (Molecular Devices), filtered at 10 kHz and sampled at 50 kHz. Patch pipettes (3–6 MΩ) were filled with an internal solution containing 130 mM KCl, 10 mM HEPES, 5 mM EGTA, 1 mM MgCl₂, and 5 mM Mg-ATP (pH 7.3 adjusted with KOH; ~295 mOsm/L). The external bath solution contained 138 mM NaCl, 4 mM KCl, 10 mM HEPES, 10 mM glucose, 2 mM CaCl₂, 1 mM MgCl₂, and 0.33 mM NaH₂PO₄·2H₂O (pH 7.3 adjusted with NaOH; ~310 mOsm/L). Series resistance (10–30 MΩ) was compensated by ~60–80%, and recordings were discarded if series resistance varied by more than 20% during the experiment. Action potentials were evoked by 4-ms depolarizing current injections (500 pA) applied at 1 Hz. Data were acquired and analyzed using pClamp 10.4 (Molecular Devices). The maximal upstroke velocity (V_max) was determined as the peak of the first derivative of the membrane voltage (dV/dt) during the depolarization phase. APD₅₀ was measured as the time from the overshoot to 50% repolarization.

Heart rate recording in zebrafish

To determine the heart rate of zebrafish, 2 dpf embryos were loaded in a recording chamber filled with E3 solution at the desired stage. Heart rate was calculated by counting the number of sequential contractions in 30 s intervals under a dissecting microscope (S8APO; Leica)⁴³. The gender of the embryos was not screened during the experiment. All zebrafish in this study were raised and handled according to a zebrafish protocol (lMM-XionguW-3) approved by the Institutional Animal Care and Use Committee at Peking University, which is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.

In order to detect the effects of compounds on zebrafish heart rate under different illumination, the control group was without treatment, and the quinidine and azo-Q2a were prepared with E3 buffer at the applied concentrations. Then, zebrafish larvae were exposed to the compounds by soaking and incubated in a 28.5 °C incubator in the dark for 1.5 h under perfusion. Next, the heart rate labeled darkness was recorded as described above. After that, giving to each group respectively 365 nm light (4500 Lux) or 365 nm after a subsequent illumination with 480 nm light (180000 Lux), the duration of each illumination is 5 minutes. Finally, we recorded the heart rate of groups labeled 365 nm or 365 nm + 480 nm after placing the larvae in the dark for 30 min after light exposure.

To further examine the therapeutic potential of azo-Q2a against Na_V1.5-related arrhythmia, a tachycardia model was established in 3–4 dpf zebrafish larvae by incubation with 10 μM isoproterenol (ISO) in Hanks’ solution for 100 min in darkness. Following ISO treatment, larvae were exposed to 10 μM cis-azo-Q2a, 10 μM trans-azo-Q2a, or 100 μM quinidine under 365 nm illumination for 5 min or kept in darkness. Heart rate was measured after an additional 30 minutes of darkness.

Transient co-expression of human Na_V1.5S and β1

The methods for transient co-expression, protein purification, and cryo-EM analysis were conducted following a standard protocol⁴⁴. The optimized coding DNAs for human Na_V1.5 (Uniprot: Q14524) were cloned into the pEG BacMam vector with twin Strep-tag and FLAG tag in tandem at the amino terminus, while β1 (Uniprot: Q07699) was cloned into the pCAG vector without an affinity tag^45,46. To optimize the protein behavior, carboxyl terminal domain (1895-2016) and linker between repeat Ⅰ-Ⅱ (461-657) and Ⅱ-Ⅲ (1066-1187) of Na_V1.5, named Na_V15S, were deleted based on the truncated construct rNa_V1.5³². HEK293F cells (Invitrogen) were cultured in SMM 293T-I medium (Sino Biological Inc.) under 5 % CO₂ in a Multitron-Pro shaker (Infors, 130 r.p.m.) at 37 °C. When cell density reached 1.8 ~ 2.2 × 10⁶ cells/ml, 2.0 mg plasmids (1.5 mg Na_V1.5S and 0.5 mg β1) and 4 mg of 40-kDa linear polyethyleneimines (PEI, Yeasen) were mixed up in 15 mL fresh medium and pre-incubated for 15- 30 min before adding to 1 liter cell culture. In addition, 10 mM sodium butyrate was added to the cell culture. Transfected cells were cultured for 48 h before harvesting.

Protein purification of Na_V1.5S and β1

14 liter transfected cells were harvested by centrifugation at 1431 g and resuspended in the lysis buffer containing 25 mM Tris-HCl (pH 7.5) and 150 mM NaCl. The suspension was supplemented with 1% (w/v) n-dodecyl-β-D-maltopyranoside (DDM, Anatrace), 0.1 % (w/v) cholesteryl hemisuccinate Tris salt (CHS, Anatrace), and protease inhibitor cocktail containing 2 mM phenylmethylsulfonyl fluoride (PMSF), 6.5 μg/ml aprotinin, 3.5 μg/ml pepstatin, and 25 μg/ml leupeptin. After incubation at 4 °C for 2 h, the cell lysate was centrifuged at 16751 g for 1 h, and the supernatant was applied to anti-Flag M2 affinity gel (Sigma) at 4 °C. After flow through by gravity, the resin was rinsed four times with the Wash buffer (25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 0.02% (w/v) glycol-diosgenin (GDN, Anatrace), and the protease inhibitor cocktail). Protein was eluted with the wash buffer plus 200 μg/mL FLAG peptide (GL.biochem). The eluent was then applied to Strep-Tactin Sepharose (IBA Lifesciences). The purification protocol was similar to the previous steps except for the elution buffer, which was Wash buffer plus 2.5 mM D-Desthiobiotin (IBA Lifesciences). The eluent was then concentrated via 100-kDa cutoff Centricon (Millipore) and further purified with size exclusion chromatography (Superose-6 Increase 10/300 column, GE Healthcare) in Wash buffer. Peak fractions were pooled and concentrated to ∼30 μL. Then, 1 mM azo-QD (effectiveness 82 %) was added to the concentrated sample 30 minutes before cryo sample preparation.

Cryo-EM data acquisition

4 μl aliquots of concentrated Na_V1.5S-azo-Q2a complex were applied to Quantifoil 300 mesh R1.2/1.3 Au grids which were glow-discharged for 35 s at medium RF level of Plasma Cleaner PDC-32G (Harrick). Then grids were blotted from both sides for 4 s at 8 °C and 100 % humidity and plunge-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Prepared grids were subsequently transferred to a Titan Krios electron microscope (Thermo Fisher) operating at 300 kV and equipped with Cs corrector, Gatan K3 Summit detector, and GIF Quantum energy filter. A total of 9,406 movie stacks were automatically collected, using AutoEMation⁴⁷ with a slit width of 20 eV on the energy filter and a preset defocus range from −1.8 µm to −1.5 µm in super-resolution mode at a nominal magnification of 64,000×. Each stack was exposed for 2.56 s with the exposure time of 0.08 s per frame, resulting in a total of 32 frames per stack. The total dose rate was 50 e^-/Ų for each stack. The stacks were motion corrected with MotionCor2⁴⁸ and binned 2-fold, resulting in a pixel size of 1.0979 Å/pixel. In addition, dose weighting was performed⁴⁹. The defocus values were estimated with Gctf⁵⁰.

Image processing

The data processing procedure was nearly identical to the one we previously reported⁵¹. In brief, a total of 6,525,104 particles were automatically picked in cryoSPARC⁵². 2D classification identified 614,955 good particles that were subsequently applied to Hetero refinement and non-Uniform refinement. Finally, a 3D EM map with an overall resolution of 3.03 Å was generated using 287,376 particles. The resolution was estimated with the gold-standard Fourier shell correlation 0.143 criterion⁵³ with high-resolution noise substitution⁵⁴.

Model building and structure refinement

The initial model of Na_V1.5S was based on the coordinates of human Na_V1.5-quinidine (PDB:6LQA)³¹, and the model building and structure refinement processes were the same as we used before⁵¹. In brief, the coordinate of Na_V15-quinidine was fitted into the EM map by CHIMERA (Detailed workflow was shown in Supplementary Fig. 9b)⁵⁵. And every residue was manually checked in COOT⁵⁶ carefully. The chemical properties of amino acids were considered during model building. The cis-azo-Q2a was fitted into the EM map. Structure refinement was performed using phenix.real_space_refine application in PHENIX⁵⁷ real space with secondary structure and geometry restraints. Over-fitting of the overall model was monitored by refining the model in one of the two independent maps from the gold-standard refinement approach and testing the refined model against the other map⁵⁸. Statistics of the map reconstruction and model refinement can be found in Supplementary Table 6.

Molecular docking simulation

The Cryo-EM structure of the Na_V1.5–cis-azo-Q2a complex was refined in Schrödinger Maestro v13.5, followed by restrained minimization using the Impref module of Impact with the OPLS4 force field, in which hydrogen atoms were optimized while allowing limited heavy-atom relaxation^59–61. The trans-azo-Q2a ligand was prepared in LigPrep to generate low-energy, all-atom conformers, with ionization states enumerated at pH 6.5 ± 1.0 and minimized using OPLS4⁶¹.

A receptor grid was defined from the cis-azo-Q2a binding pocket, and trans-azo-Q2a was docked into Na_V1.5 in both unconstrained and pose-constrained modes, the latter referencing the cis-azo-Q2a orientation. Docking was performed using Glide in extra-precision (XP) mode, and poses were ranked by GlideScore and evaluated by MM/GBSA to estimate relative binding free energies⁶².

Data analysis and Statistics

For in vitro experiments, the cells were evenly suspended and then randomly distributed in each well tested. For in vivo experiments, the zebrafish were distributed into various treatment groups randomly. Statistical analyses were performed using GraphPad Prism 10 (GraphPad Software). Before statistical analysis, variation within each group of data and the assumptions of the tests were checked. Comparisons between two independent groups were made using an unpaired Student’s two-tailed t-test. Comparisons among nonlinear fitted values were made using the extra sum-of-squares F test. Comparisons among three or more groups were made using ordinary one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001. Values of IC₅₀ are presented as mean (95 % CI (Profile likelihood)). Other data are presented as mean ± SEM.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Z.H., H.L., S.L., and W.G. conceived and designed the experiments. S.L. and W.W. carried out the patch-clamping recordings and constructed all the mutations. H.S. recorded the cardiac action potentials. W.G. synthesized compounds and collected photochemical data. S.L., W.W., J.H., and J.X. collected heart rate data in zebrafish. Z.L., J.L., H.W., and N.Y. performed all the experiments of cryo-EM data. S.L., W.G., Z.L., M.Y., X.T., Z.H., and H.L. wrote the paper. X.T. is the lead corresponding author. All authors reviewed and revised the paper.

Peer review

Peer review information

Nature Communications thanks Daohua Jiang, Jerome Montnach, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data supporting the findings of this study are available within the paper and its supplementary information files. Source data are provided with this paper. All other data supporting the findings of this study are available from the corresponding author on reasonable request. The atomic coordinate of Na_V1.5S-azo-Q2a has been deposited in the Protein Data Bank under the accession code 9V3S. The EM map of Na_V1.5S-azo-Q2a has been deposited in the Electron Microscopy Data Bank with the accession code EMD-64755. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70305-6.