Heart rate control by listening to light

Abstract

The auditory-cardiac interplay enables sound perception to modulate cardiac rhythm and homeostasis. However, its underlying regulatory mechanism and therapeutic potential remain largely underexplored, due to fundamental limitations in acoustic stimulation precision. Here we present an opto-mechano-cardiac coupling strategy that bypasses the propagation constraints of sound wave to enable precise acoustic modulation of cardiac function. Utilizing dynamically reconfigurable optical traps (1064 nm, 100 kHz refresh rate), we translated musical rhythms into programmable otolith oscillation patterns, thus establishing deterministic input-output relationships between mechanical stimulation and cardiac responses. We found that hindbrain areas, primarily the vagal motor area, mediate this opto-mechano-cardiac modulation, through adrenergic activation and a shift in the balance between sympathetic and parasympathetic activity. Leveraging real-time rhythm transcoding, we precisely quantified the modulation capabilities of auditory stimuli with distinct spectral composition on cardiac rhythm, and successfully rescued the drug-induced pathological cardiac rhythm disorders. By bridging optical physics and autonomic cardiac physiology via optomechanical otolith oscillation, this work establishes a mechanistic framework to explore sensory-cardiac modulation and personalized arrhythmia management, paving the way for advancing music therapy from empirical practice to quantitative biomedicine.

The auditory-cardiac interplay allows sound to influence cardiac rhythm, but precise control is limited by acoustic wave propagation. Here, the authors present an opto-mechano-cardiac coupling strategy that uses optical traps to convert music rhythm into programmable otolith oscillations, identifying key neural pathways and successfully rescuing drug-induced arrhythmias.

Introduction

When listening to preferred music, individuals frequently experience perceptual alignment, in which cardiac rhythm is modulated by musical tempi^1,2, although the extent and direction of this modulation vary significantly, depending on musical training, emotional engagement and autonomic state^3,4. Such observations have inspired systematic research into the intricate interplay between auditory perception and cardiac rhythm, prompting the development of non-invasive music therapy for arrhythmia^5–7. Through curated musical selection, personalized interventions can be tailored to individual pathophysiological profiles to regulate heart rate, and consequently ameliorate cardiac function via autonomic nervous system modulation^8–10. Nevertheless, the biophysical mechanisms underlying sound-cardiac coupling remain obscure, largely due to the diffusive propagation of acoustic waves and the multifaceted nature of auditory activation^11,12. In particular, the reliance on airborne pressure waves fundamentally limits the spatiotemporal precision of controlling mechanosensory inputs and decoding organ-specific neural pathways, thereby hindering the efficient development of personalized arrhythmia interventions^13,14.

Optical mimicry of auditory stimulation offers a promising avenue to overcome these barriers. For instance, the gradient force of a tightly focused laser beam can induce controlled mechanical oscillations in biological microstructures^15–17. The vertebrate otolith system¹⁸, an evolutionarily conserved inertial guidance module that transduces mechanical deflections into auditory activation via hair cells^19–21, serves as an ideal photonic interface. Encouragingly, non-contact manipulation of otoliths has been achieved by using optical tweezers^22,23, demonstrating the feasibility of modulating the auditory system through optical technologies. Here, we introduce an opto-mechano-cardiac coupling strategy to modulate cardiac rhythm via optomechanical otolith oscillation. This study was performed on larval zebrafish model, capitalizing on its excellent optical transparency for in vivo optical manipulation and real-time cardiac monitoring^24–26, flexible genetic accessibility for probing neural activity^27,28 and dissecting sensory-cardiac interactions^29,30, and rapid morphogenesis for characterizing the efficacy of optomechanical cardiac regulation across developmental milestones^31–33. By applying piconewton-scale forces with submicron spatial precision, dynamically reconfigurable optical traps permit selective stimulation of discrete saccular (Sac) or utricular (Ut) pathways. This capability allows us to characterize the precise tuning of cardiac response to various oscillation parameters, including displacement amplitude, oscillation frequency and angular directionality. Crucially, this approach enables real-time conversion of musical waveforms into tailored oscillation profiles, establishing the physical basis for photonic-music transduction and allowing quantitative exploration of mechano-electrophysiological coupling in auditory-cardiac regulation. By systematically mapping distinct otolith oscillations to auditory-cardiac coupling, this strategy bypasses the barriers of acoustic propagation and establishes a mechanistic framework for advancing music-based personalized arrhythmia therapies.

Results

Schematic sketch of optomechanical otolith oscillation for auditory regulation

Figure 1 illustrates the auditory perception and cardiac modulation via optomechanical otolith oscillation. Two optical traps at a wavelength of 1064 nm are alternately scanned on opposite sides of the target otolith (Fig. 1b) at high speed (maximum refresh rate: 100 kHz), with the aim of driving a periodic positional oscillation. Notably, optomechanical otolith oscillation activates hair cells by inducing a dynamic deflection of the connected ciliary bundles anchored to them^34,35. It thus mimics the physiologically authentic otolith oscillations elicited by acoustic stimuli and initiates neural signal propagation through the auditory afferents. In other words, the organism perceives photon stimuli through auditory neural pathways rather than visual systems. Meanwhile, synchronized heartbeat responses are monitored during optomechanical oscillation (Fig. 1c), with modulation magnitudes dependent on specific oscillation patterns. Compared with conventional acoustic stimuli, this optical strategy enables individual otolith targeting with sub-micrometer precision in positioning and programmable waveform generation, thus permitting quantitative investigation of music-cardiac interactions and advancing music therapy from empirical practice to personalized arrhythmia management.

Fig. 1. Schematic sketch of opto-mechano-cardiac modulation and material characterization.

a Schematic sketch of heart rate control by optomechanical otolith oscillation. Two alternating 1064 nm optical traps drive periodic otolith oscillation to mimic acoustic stimulation, activating hair cells via ciliary bundle deflection and initiating auditory afferent signaling, while synchronized heartbeat responses are monitored. b,c Transmission light microscopic image (b) and corresponding fluorescence image (c) of larval zebrafish for identifying the cardiac chamber and its spatial relationship to the otolith. The Tg(cmlc2:EGFP; flk:mCherry) double-transgenic line was employed to concurrently label the vasculature (red) and the heart (green). This image is representative of observations from n = 5 larvae.

Programmable optomechanical otolith oscillation

To validate the feasibility of programmable optomechanical otolith oscillation, multifunctional control capabilities were characterized first for optical manipulation over otolith dynamics on 6 days post-fertilization (dpf) larval zebrafish (Fig. 2a). The 6 dpf larval zebrafish was chosen because it represents a critical window where the auditory and cardiovascular systems are functionally mature^31,36, yet the larvae retain optimal optical transparency for high-resolution imaging and precise optomechanical manipulation. In this study, a 1064 nm laser was employed for optomechanical oscillation, leveraging its relatively low tissue absorption in biological specimens^37,38. After interacting with an acousto-optic deflector (AOD), the incident beam was reflected by the dichroic mirror and then focused on the target otolith via a high-NA water-immersion objective (60×, NA = 1.0), thus generating dual optical traps with sub-micrometer positioning accuracy (Supplementary Fig. 1). By reconfiguring the spatiotemporal arrangement of these traps, the otolith could undergo programmable oscillations with controlled displacement amplitudes, oscillation frequency and angular deflections, thus enabling precise encoding of target oscillation patterns corresponding to specific musical waveforms.

Fig. 2. Optomechanical otolith oscillation for encoding musical patterns.

a Schematic sketch of optomechanical otolith oscillation. The 1064 nm beam is steered by an AOD and focused by a high-NA objective to generate dual optical traps for controlled otolith oscillation. The inset indicates the displacement amplitude and directional angle θ. b Optical microscopic images for oscillating the utricle otolith along different directions (P = 120 mW and f = 5 Hz). The induced displacement was verified by quantifying the shift in the otolith’s intensity profile before and after optomechanical oscillation. This image is representative of observations from n = 5 larvae. c The recorded kymograph shows the motion of utricle otolith under various oscillation frequencies (P = 100 mW). d The calculated oscillation frequency of utricle otolith exhibited a linear correlation with the oscillation frequency of optical traps (P = 600 mW). e The calculated displacement amplitude of utricle otolith as a function of oscillation frequency of optical traps (P = 180 mW). Data are presented as median values with interquartile ranges. Center line, median; box limits, upper (75th percentile) and lower (25th percentile) quartiles; whiskers, minimum and maximum values. f The displacement amplitude of otoliths as a function of exerted laser power at the oscillation frequency of 5 Hz. g Quantitative characterization of oscillation flexibility towards different directions (P = 180 mW and f = 5 Hz). All experiments were performed on 6-dpf larval zebrafish. Results are presented as mean ± SD (n = 5 individual larval zebrafish).

As previously shown, the otolith was mechanically coupled to hair cells via kinocilia bundles^39,40, forming the mechanosensitive interface for auditory transduction (Supplementary Fig. 2). To induce optomechanical oscillations, a pair of optical traps (P = 120 mW) were positioned on opposite sides of the otolith and alternately turned on (Figs. 2b2,3), generating lateral displacements of 2.0 ± 0.05 μm (leftward) and 2.0 ± 0.05 μm (rightward) under the action of optical gradient force. Further adjustments to the trap arrangement confirmed vertical oscillation modes, achieving vertical displacements of 1.3 ± 0.08 μm (upward, Fig. 2b4) and 1.1 ± 0.07 μm (downward, Fig. 2b5). The rapid reconfiguration of oscillation pattern allowed seamless transitions between oscillation axes, mimicking natural multidirectional sound-induced otolith motions¹³. Notably, the difference between vertical and horizontal displacements could be attributed to the anatomical arrangement in which the kinocilial bundles of hair cells are anchored to the otolith at a distinct orientation. When the otolith is moved horizontally (laterally), it primarily induces a shearing force that causes a lateral deflection (or bending) of the ciliary bundles, while a vertical oscillation, in contrast, primarily exerts a compressional/elongational force along the major axis of the bundles. This anisotropy is a fundamental feature that enables the organism to discriminate the directionality of acoustic stimuli, demonstrating the ability of optomechanical oscillations to mimic physiologically distinct mechanical stimuli and to study directional auditory processing. Upon deactivating optomechanical oscillation, the otolith returned to its equilibrium position within 1 s (Fig. 2b6), confirming the non-invasive and reversible nature of the developed optomechanical oscillation platform (Movie S1).

Fig. 3. Heart rate control by optomechanical otolith oscillation.

a A 6-dpf zebrafish heart accelerates its beat when the otolith experiences optomechanical oscillation from 3.3 to 9.5 s and recovers afterward. The kymograph shows the motion of the heart wall along the horizontal direction. b Quantitative comparison of ΔHR with optical traps exerted on various locations. c Relationship characterization of ΔHR on otolith oscillation (Frequency: 10 Hz; Amplitude: 1 μm; Direction: 0°). d–f The calculated ΔHR as a function of displacement amplitude (d) (Frequency: 5 Hz; Direction: 0°), oscillation frequency (e) (Amplitude: 1 μm; Direction: 0°) and angular direction (f) (Frequency: 10 Hz; Amplitude: 1 μm). g The calculated ΔHR as a function of otolith subtypes (Frequency: 10 Hz; Amplitude: 1 μm; Direction: 0°). (h) Comparison of cardiac modulation efficacy between ipsilateral and contralateral activation of homologous otoliths (Frequency: 10 Hz; Amplitude: 1.5 μm; Direction: 0°). Results are presented as mean ± SD (n = 5 individual larval zebrafish). Inter-group difference was tested using one-way ANOVA with Bonferroni correction. P values < 0.05 were considered statistically significant.

Building upon this foundation, the operational flexibility was systematically quantified for optomechanical otolith oscillation. By programming the spatial arrangement and scanning frequency of the beam foci, the oscillation parameters, including displacement amplitude, oscillation frequency and angular direction, can be finely tuned for the target otolith in real time. As demonstrated in Fig. 2c, real-time programming of optical traps enabled flexible modulation of oscillation frequency, as exemplified at 5, 10, and 15 Hz (Supplementary Fig. 3). Further validation confirmed precise synchronization between the optical trap scanning and the resultant otolith oscillations across the frequency range of 0 ∼ 500 Hz (R² > 0.99 across 5 trials, Fig. 2d, Supplementary Fig. 4 and Movie S2). Notably, the integrated AOD has a maximum switching rate of 100 kHz, thus enabling optomechanical otolith oscillation that can, in theory, be performed at frequencies up to 50,000 Hz. This range encompasses the hearing ranges of both zebrafish (approximately 100– 4000 Hz) and humans (20–20,000 Hz), demonstrating the optomechanical platform’s capacity to encode a range of musical frequency components. Under constant laser power (P = 180 mW), the displacement amplitudes were measured across five individual larval zebrafish and exhibited an inverse relationship with oscillation frequency (Fig. 2e and Supplementary Fig. 5), decreasing by 60 ± 8% as frequency increased from 5 to 40 Hz. This amplitude-frequency coupling likely arises from the viscoelastic properties of connected ciliary bundles, under which higher-frequency switching reduces the effective force duration per oscillation cycle. Conversely, at fixed frequencies (e.g., 5 Hz), incremental increases in laser power produced proportional amplitude enhancements (Fig. 2f). Comparative analysis revealed that utricular otolith exhibited greater displacements than saccular otolith under identical trapping power, which might be attributable to the anatomical distinctions such as otolith mass and ciliary bundle length⁴¹. Further analysis demonstrated anisotropic responsiveness on the oscillation direction, and maximum displacements occurred along the anteroposterior axis (0°–180° horizontal plane, Fig. 2g and Supplementary Table 1), which aligns with zebrafish biomechanical sensitivity to head-tail accelerations. Both the utricle and saccular otoliths exhibited equivalent directional preferences, suggesting conserved auditory mechanical alignment across inner-ear compartments. This bidirectional control paradigm demonstrated submicrometer spatial resolution and millisecond-scale temporal fidelity, essential for mimicking physiological oscillation ranges observed during auditory processing.

Heart rate control by optomechanical otolith oscillation

Following the establishment of programmable optomechanical oscillation, its capacity to regulate heart rate was systematically evaluated in 6-dpf larval zebrafish. Under resting conditions, zebrafish exhibited a baseline heart rate of 2.0 Hz. Upon initiating optomechanical oscillation at t = 3.3 s (Frequency: 50 Hz; Amplitude: 1 μm; Direction: 0°), the heart rate progressively increased to 2.7 Hz after oscillation duration of 1.7 s, and then gradually returned to baseline levels upon oscillation cessation at t = 9.5 s (Fig. 3a). Control experiments were introduced to exclude confounding effects from mechanical perturbation or laser-induced photothermal stress, including applying the same optical traps to five distinct regions: exterior of the zebrafish, dorsal body surface, hair cell on the lateral line, non-sensory inner ear areas, and the otolith itself, together with a no-trap control group. Crucially, the locations of exerted optical traps were anatomically consistent across all experimental conditions, with the distinct locations labeled in Supplementary Fig. 6. To quantify cardiac responses with enhanced precision, heart rate change (ΔHR), defined as the difference of the heart rate after optomechanical otolith oscillation from that before the oscillation, was measured to reveal the oscillation-dependent autonomic modulation. The results indicated that non-target stimulation (dorsal body or non-sensory inner ear regions) induced marginal heart rate fluctuations, in contrast to a maximal 50% heart rate increase upon otolith-targeted oscillation (Fig. 3b). Notably, optomechanical oscillation of hair cells on the lateral line failed to elicit observable heart rate change (Supplementary Fig. 7), which might be attributed to the functional divergence and the stimulation efficacy difference for the hair cells between inner ear and lateral line. Meanwhile, optomechanical otolith oscillation synchronously engages a broad population of hair cells by collectively deflecting their ciliary bundles, while stimulating a single or a few hair cells in the lateral line provides a relatively localized and weak afferent input, which is probably insufficient to recruit the autonomic pathways involved in heart rate modulation. Static otolith trapping only evoked a minor cardiac response (ΔHR < 10%), whereas controlled oscillation elicited a significant heart rate elevation (Fig. 3c and Movie S3). These findings confirm that cardiac modulation arises primarily from mechanosensory transduction during otolith oscillation, rather than photothermal artifacts.

Then, the parametric relationships between artificial optic-mechano-acoustic stimuli and autonomic cardiac responses were quantified. The ΔHR exhibited a linear increase with displacement amplitude (Fig. 3d), which likely reflects graded mechanotransduction: smaller oscillations induced weaker deflection of ciliary bundles, generating slight neural signals that minimally alter ΔHR. Conversely, larger otolith amplitudes, corresponding to greater deflection of hair cell bundles, are expected to elicit stronger depolarization in the afferent neurons⁴², which is consistent with the enhanced neural calcium signals within the central brain regions (Supplementary Fig. 8). Compared to the displacement amplitude, frequency-dependent effects were subtle (Fig. 3e). At matched amplitudes (1 μm), ΔHR increased marginally across the oscillation of 10 Hz (37 ± 3%), 50 Hz (41 ± 5%), and 100 Hz (46 ± 3%). This limited frequency discrimination might arise from the restricted diversity of ciliary bundles directly coupled to the oscillated otolith, unlike the broader frequency-tuned hair cell populations in the whole auditory system. Further analysis revealed significant dependence of cardiac modulation efficiency on the oscillation direction. In contrast to other orientations (45°, 90° and 135°), horizontal oscillations (0° alignment) elicited maximal heart rate changes of 35 ± 3% (Fig. 3f). This direction preference correlates with the anatomical polarization of hair cells, where maximal mechanotransduction signals are generated when cilia are deflected along their intrinsic sensitivity axis⁴².

Given the functional specialization of utricular and saccular otoliths, their cardiac regulatory capacities were quantified and compared under identical optomechanical oscillation parameters. The results indicated the saccular otolith elicited a greater heart rate elevation compared to the utricular otolith (Fig. 3g), aligning with its evolutionary role in encoding ambient acoustic cues than vestibular signals⁴³. Equivalent cardiac modulation efficacy was observed for ipsilateral versus contralateral activation of homologous otoliths (Fig. 3h), confirming symmetric auditory neural pathway engagement. Moreover, simultaneous dual-otolith oscillation revealed additive cardiac effects, i.e., the summed ΔHR values from individual utricular and saccular oscillation closely matched ΔHR during concurrent activation (Supplementary Fig. 9). This linear summation suggests parallel neural coding of artificial mechano-acoustic stimuli from distinct otolith subtypes, with limited central integration saturation. These findings inspire subtype-specific and combinatorially tunable cardiac regulation through selective otolith oscillation, providing a platform to dissect hierarchical integration of multisensory inputs in autonomic control circuits.

Neural signaling pathways behind optomechanical cardiac modulation

The activity of brain regions upon the optomechanical cardiac modulation was examined by brain-wide neural imaging on transgenic zebrafish larvae expressing calcium sensor (GCaMP6f) in all neurons (huc: GCaMP6f; nacre)⁴⁴ (Fig. 4a and Supplementary Fig. 8). For calcium imaging of neural activity, the zebrafish was mounted in a dorsal recumbency position with the brain surface titled downward (Supplementary Fig. 10), as optimized for brain imaging using an inverted microscope, instead of the lateral position for simultaneous otolith manipulation and cardiac imaging in Fig. 1b. As optomechanical oscillation introduced at t = 2 s, calcium transients were enhanced in the vagal motor nucleus, the torus semicircularis (TS), and the thalamus regions (Fig. 4b, c and Supplementary Figs. 11, 12). Among them, the torus semicricularis is known as the auditory sensing region⁴⁵, and the vagal lobe is known to harbor the neurons receiving and sending the autonomic innervations of the peripheral body organs including heart^46,47. Besides, the medulla oblongata also harbors norepinephrinergic neurons involved in broad neuromodulation. These activated regions confirmed that artificial otolith oscillations engage canonical sound-responsive neural circuits. Furthermore, the latency difference between the hindbrain activation (0.84 ± 0.02 s) and the heart rate elevation (1.68 ± 0.03 s) was consistent with the role in the neural mediation of cardiac responses⁴⁸.

Fig. 4. Neural signaling pathways behind optomechanical cardiac modulation.

a,b Calcium fluorescence imaging of zebrafish larvae brains (a) and corresponding time-resolved color mapping of neural activity (b) during optomechanical otolith oscillation. TS: torus semicircularis. The color coding represents different time points during the calcium imaging recording following the onset of otolith oscillation. Data are from n = 3 larval zebrafish. c Quantification of calcium response onset latencies across brain regions during optomechanical otolith oscillation (Frequency: 10 Hz; Amplitude: 1 μm; Direction: 0°). Response onset latency was determined as the time point of maximal slope in the rising phase of the calcium transient, extracted via K-means clustering-based trajectory analysis of normalized calcium signals. VMN and TS represent vagal motor nucleus and torus semicircularis, respectively. Data are presented as mean ± SD (n = 3 larval zebrafish). d Schematic sketch of depicting the pharmacological blockade approach to explore neural pathways during optomechanical cardiac regulation. e Quantitative comparison of ΔHR during optomechanical otolith oscillation (Frequency: 10 Hz; Amplitude: 1 μm; Direction: 0°) while under different pharmacological treatments. The used drug concentration was 2 mM, 6 mM and 4 mM for MS222, Propranolol and Atropine, respectively. f–h Dose-dependent variations in hear rate during optomechanical otolith oscillation (Frequency: 10 Hz; Amplitude: 1 μm; Direction: 0°) while under increased concentration of Propranolol (f), Atropine (g) and MS222 (h). i Schematic diagram portraying neural circuit mechanisms behind optomechanical cardiac regulation. Significance of difference was tested using one-way ANOVA with Bonferroni correction (n = 6 larval zebrafish). P values < 0.05 were considered statistically significant.

The vagal motor nucleus sends parasympathetic nerves, which control cardiac functioning via cholinergic transmission. Furthermore, norepinephrinergic innervation from medulla oblongata and other brainstem regions modulates cardiac activity in “fight or flight” decisions⁴⁹. Based on these, their involvements were further confirmed with a pharmacological blocking experiment, in which the larval zebrafish were incubated in propranolol (2–6 mM, β-adrenergic antagonist), and atropine (2–6 mM, muscarinic cholinergic antagonist) solutions (Fig. 4d). After that, they were subjected to optomechanical otolith oscillation while drug-induced changes in heart rate were measured. Following otolith oscillation in larvae zebrafish, propranolol-treated groups exhibited dose-dependent ΔHR suppression (Fig. 4e, f). This aligns with the β-adrenergic identity of the principal sympathetic pathway for heart rate acceleration, where propranolol competitively inhibits norepinephrine binding to cardiac β-receptors, thereby attenuating sympathetic-driven heart rate elevation⁵⁰. In contrast, atropine pretreatment induced dose-dependent ΔHR amplification (Fig. 4e, g). This paradoxical enhancement stems from atropine’s blockage of cardiac muscarinic receptors, which will prevent acetylcholine-mediated potassium efflux and membrane hyperpolarization, thereby suppressing parasympathetic braking on heart rate⁵¹. Notably, MS222 anesthetic controls showed no significant ΔHR modulation (Fig. 4h). Taken together, the suppressive effect of propranolol provides conclusive evidence that adrenergic activation contributes to the heart rate increase during optomechanical otolith oscillation. However, the amplifying effect of atropine does not necessarily indicate an active increase in cholinergic efferent activity during optomechanical oscillation, as baseline parasympathetic activity is always present. Therefore, pharmacological data support a model of sympathetic dominance but leave unresolved whether the cholinergic contribution reflects the removal of ongoing baseline parasympathetic tone or an active increase in cholinergic efferent activity during optomechanical oscillation. Further investigations are required to independently distinguish these two possible mechanisms underlying the cholinergic contribution.

Based on these findings, a neuroanatomical model can be established for optomechanical cardiac modulation (Fig. 4i). When otolith oscillations deflect ciliary bundles on hair cells, mechanotransduction initiates and generates an action potential in the vestibulocochlear nerve. These signals converge at the thalamus which act as a sensory gatekeeper^52,53 and are relayed to the hindbrain areas. The final efferent pathway involves adrenergic efferent sending from the medulla oblongata to the sinoatrial node (SAN), which releases norepinephrine to potentiate pacemaker activity. This hierarchical circuit, i.e., from otolith mechanotransduction to vagal motor nucleus integration and autonomic efferent signaling, resembles the neurocardiac cascade triggered by natural acoustic stressors, validating synthetic optomechanical oscillation as a biomimetic tool for dissecting autonomic regulation.

Musical encoding and antiarrhythmic therapy by optomechanical otolith oscillation

Building upon programmable otolith oscillation across a much broader spectrum (i.e., 0∼50,000 Hz), the proposed optomechanical strategy might allow a real-time musical encoding, i, translating various auditory stimuli with distinct spectral compositions into specific otolith oscillation patterns. To bridge artificial optical control with realistic sound stimuli, the fundamental frequencies of Do (C4), Re (D4), Mi (E4), Fa (F4), Sol (G4), La (A4) and Ti (B4) were mapped with 261.6 Hz, 293.7 Hz, 329.6 Hz, 349.2 Hz, 392.0 Hz, 440.0 Hz and 493.9 Hz, respectively. Based on this authentic frequency set, two kinds of musical stimuli were quantitatively defined by their distinct spectral composition and temporal dynamics (Fig. 5a): 1) the “exciting music” was characterized by broad spectral coverage and rapid transitions, i.e., the stimulus pattern jumps across the full set of frequencies (261.6∼ 493.9 Hz) with a fast rhythm of 60 beat per minute (bpm), meaning the music pattern changes once per second. 2). The “soft music” was characterized by restricted spectral range and smooth transitions, i.e., the stimulus pattern is confined to a narrower mid-frequency band (i.e., Mi ∼ Sol: 329.6∼ 392.0 Hz) with a slow rhythm of 20 bpm, meaning the music pattern changes only once every three seconds. The system’s performance in encoding musical stimuli was validated by measuring the otolith’s oscillation frequency in response to the “exciting” and “soft” music patterns (Fig. 5b). Notably, the measured frequencies (represented by circular symbols for soft music and star symbols for exciting music) closely matched the target frequencies of musical scale, confirming the system’s fidelity in translating complex acoustic information into mechanical stimuli.

Fig. 5. Antiarrhythmic therapy by optomechanical otolith oscillation.

a Differentiating the spectral composition and temporal dynamics of two auditory stimuli. b Real-time encoding of musical rhythm signatures into specific optomechanical oscillation patterns with color mapping denoting the frequency spectrum allocation of musical components. The experimentally measured frequency (circular symbols for soft music and star symbols for exciting music) is overlaid with the designed musical rhythm (n = 5 larval zebrafish each group). c Comparative heart rate analysis under music-converted optomechanical otolith oscillation. Results are presented as mean ± SD (n = 5 larval zebrafish each group). Significance of difference was tested using one-way ANOVA with Bonferroni correction. d,e Representative kymograph trace d and comparative heart rate analysis (e) in exciting-music based bradycardia rescue. Results are presented as mean ± SD (n = 7 individual larval zebrafish). Data was analyzed using one-way ANOVA with Bonferroni correction. (f) Representative kymograph analysis and AV ratio quantification pre-/post-optomechanical otolith oscillation. g Efficacy characterization of exciting music therapy in rescuing AV block. Bar graph quantifying the number of larval zebrafish exhibiting AV block before and after exciting music-based optomechanical stimulation (n = 7 individual larval zebrafish). h Representative kymograph trace depicting sinus pause rescue via optomechanical otolith oscillation. i Efficiency characterization of exciting music therapy in rescuing sinus pauses. Bar graph quantifying the number of larval zebrafish exhibiting sinus pause before and after exciting music-based optomechanical stimulation (n = 7 individual larval zebrafish). P values < 0.05 were considered statistically significant.

Leveraging this musical encoding capacity, the modulation capabilities were quantitatively characterized for musical patterns on cardiac rhythm. Under resting conditions (baseline heart rate: 2.5 ± 0.2 Hz), soft music induced a mild heart rate increase to 3.1 ± 0.2 Hz, while exciting music triggered a significant elevation to 3.65 ± 0.25 Hz (Fig. 5c). Population-level validation across five larval zebrafish demonstrated a consistent trend, for which the differential cardiac responses are directly correlated with the spectral composition of the encoded music stimuli. This music-to-mechanoacoustic encoding strategy decouples physical stimulus parameters from psychological perception, revealing that spectral composition-specific cardiac effects are driven by mechanical signal dynamics. By mapping tempos, dynamic ranges, and rhythmic regularity to optomechanical oscillation parameters, this approach might establish a quantitative framework for the development of bio-inspired auditory therapies for various arrhythmia diseases.

Aiming at this goal, its therapeutic potential was evaluated in different arrhythmia models. Based on its heart rate-accelerating effect, the proposed optomechanical oscillation was first applied to treat bradycardia, which was induced by the anticancer drug doxorubicin. Notably, doxorubicin is a widely used chemotherapeutic agent whose clinical application is severely limited by dose-dependent cardiotoxicity, including bradycardia^54,55. As indicated in Fig. 5d, doxorubicin-treated larval zebrafish exhibited significantly reduced heart rates (1.50 ± 0.05 Hz) compared to untreated controls (2.30 ± 0.05 Hz), confirming successful bradycardia induction. To treat doxorubicin-induced bradycardia, the exciting music-encoded optomechanical oscillation was applied to the saccular otolith, which was selected due to the superior heart-rate enhancement of exciting music relative to the soft music (Fig. 5c) and the greater regulatory sensitivity of the saccular otolith relative to the utricular otolith (Fig. 3h). Following 10 minutes of exciting music stimulus, the heart rate increased significantly from the bradycardic state to a rate of 2.20 ± 0.05 Hz (Fig. 5e), demonstrating the music-driven therapeutic potential. To validate sustained heart rate elevation, an extended post-intervention monitoring was performed after the 20-minute musical therapy, to track heart rate for additional 60 min after cessation of optomechanical stimulation (Supplementary Fig. 13a). The results indicated that the heart rate at the end of 20-min treatment (i.e., 2.05 ± 0.1 Hz at t = 0 min post-treatment) exhibited a gradual decline to a stabilized rate of 1.95 ± 0.05 Hz over the subsequent 20 min (Supplementary Fig. 13b), and then maintained throughout the remaining 40-minute observation period (6 out of the 7 larvae tested, ~86% success rate). This sustained stabilization, significantly above the pre-treatment pathological level, robustly demonstrates a persistent therapeutic effect.

In addition to bradycardia symptoms, the therapeutic capability was also characterized in atrioventricular (AV) block, where the abnormal signal transduction induces asynchrony rhythms between atrium and ventricle⁵⁶. Terfenadine, a former antihistamine, was withdrawn from the market precisely due to its adverse effect of provoking AV block model of conduction abnormalities^57,58. As indicated in Fig. 5f, Terfenadine-treated larval zebrafish developed AV block characterized by 2:1 atrial-to-ventricular contraction ratios (region I), where two atrial beats preceded a single ventricular contraction (Movie S4). Application of exciting music-encoded optomechanical oscillations to saccular otolith progressively restored physiological 1:1 AV synchrony within 3 min (region II). Continued oscillation for 17 min stabilized the restored rhythm (region III, 6 of 7 embryos), which persisted in 5 responders for over 20 min post-intervention (Fig. 5g), demonstrating a durable electrophysiological remodeling effect.

Moreover, the therapeutic efficacy was further assessed in sinus pause with escape rhythms. The sinus pause was induced by using sunitinib, a receptor tyrosine kinase inhibitor, which has considerable efficacy in metastatic renal cell carcinoma⁵⁹. However, sunitinib’s clinical use is associated with various cardiac dysfunction, including hypertension, heart failure and arrhythmia^60,61. By incubating the larval zebrafish with sunitinib solution, an arrhythmia model was established which exhibited sinus pause with a mean frequency of 2 ∼ 3 events per 30 s and individual event durations ranging from 1 to 3 s (Fig. 5g and Supplementary Fig. 14). After that, the exciting music-based optomechanical oscillation was exerted onto the saccular otolith, during which the sinus node automaticity was successfully restored in 5 out of the 7 larvae tested (~71% success rate), eliminating ectopic junctional rhythms and reinstating physiological pacing (Fig. 5h and Movie S5). Post-oscillation monitoring confirmed sustained normal sinus rhythm in 5 of 7 larval zebrafish (Supplementary Fig. 14), while the remaining 2 larvae also showed significant improvement, characterized by a reduction in the frequency of sinus pauses (from 3 to 1 event per 30 s) and a shortening of the average pause duration (from 3 s to 1 s), although full restoration was not achieved.

Discussion

By engaging natural sound-evoked signaling pathways via dynamically reconfigurable optical traps, this optomechanical strategy achieves precise parameterization of stimulus dynamics. Applying piconewton-scale forces with submicron spatial precision enables isolated stimulation of saccular versus utricular pathways, a feat currently unattainable with conventional acoustics²³. These parametric characterizations validate the strategy’s capacity to emulate naturalistic otolith dynamics while achieving synthetic control beyond physiological ranges. The observed frequency-amplitude tradeoffs and directional anisotropies provide critical constraints for designing music-inspired oscillation patterns that optimize neural encoding fidelity while avoiding mechanical saturation of hair cell transduction. Crucially, the observed 0° efficacy peak suggests that horizontally aligned oscillations optimally engage the autonomic pathway by mimicking natural head-motion stimuli. These findings demonstrate that directional control of artificial otolith oscillations can fine-tune cardiac modulation intensity through spatially selective recruitment of mechanosensory circuits.

Following the parameterization of optomechanical oscillation, real-time calcium imaging in a transgenic line elucidates a direct opto-mechano-cardiac pathway functioning with millisecond-scale temporal fidelity. The spatiotemporal congruence between optomechanical oscillation and natural acoustic-evoked activation patterns demonstrates shared neural encoding mechanisms for synthetic and physiological mechano-acoustic stimuli. Further pharmacological blockade demonstrates that optomechanically evoked cardiac acceleration is mediated by adrenergic activation and a shift in the balance between sympathetic and parasympathetic activity. The receptor-level specificity confirms synthetic otolith oscillations engage evolutionarily conserved neurocardiac effector pathways identical to natural auditory-stress responses. Beyond autonomic nervous system activity, we hypothesized that hormonal secretion from the adrenal/interrenal gland might also be involved. To test this, BAPTA (1,2-bis(2-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid), a highly effective calcium chelator⁶², was delivered into the interrenal gland region of larval zebrafish via a microinjection approach. This intervention aimed to inhibit the gland’s activity by sequestering intracellular calcium, thereby reducing hormone secretion (e.g., cortisol). To rule out potential confounding effects from the microinjection procedure itself, a sham control group was included, wherein the zebrafish were injected with the same volume of physiological saline. As indicated in Supplementary Fig. 15, the heart rate modulation induced by optomechanical otolith oscillation was significantly attenuated in the BAPTA-injected group, with the ΔHR decreasing from 0.70 ± 0.05 Hz to 0.38 ± 0.05 Hz (p < 0.0001). In contrast, the sham control group exhibited no significant alteration in the cardiac response (0.70 ± 0.05 Hz vs. 0.65 ± 0.05 Hz, p = 0.1725). This significant divergence confirms the existence of a hormone-driven regulatory mechanism that operates in concert with the direct neural pathway.

This technology optimizes music therapy by encoding diverse musical styles into specific oscillation patterns and quantifying spectral composition-specific autonomic responses. In drug-induced arrhythmia models, synthetic auditory stimulation not only acutely rescues pathological rhythms but also induces lasting electrophysiological remodeling, mimicking exercise-induced chronotropic competence. The non-invasive and neuro-modulatory approach offers a safer alternative to direct cardiac pacing while achieving sustained rhythm correction, underscoring its potential for treating autonomically mediated arrhythmias. Furthermore, the integration of a hormonal secretion mechanism provides a possible explanation for the sustained therapeutic effects observed in the antiarrhythmic experiments. Specifically, the successful rescue of abnormal rhythms, which persisted even after the cessation of optical stimulation, may be attributed to optomechanical modulation inducing prolonged alterations in relevant hormone secretion, thereby maintaining normal cardiac rhythm beyond the period of direct optical intervention. To investigate this sustained therapeutic mechanism, direct β-adrenergic stimulation using isoproterenol^63,64 was evaluated in these drug-induced arrhythmia models. As shown in Supplementary Fig. 16, Isoproterenol incubation effectively rescued bradycardia in doxorubicin-treated larvae, restored 1:1 atrioventricular conduction in Terfenadine-treated larvae (5 of 7 animals), and significantly reduced Sunitinib-induced sinus pauses, with these improvements persisting 20-min after washout. Collectively, these findings suggest that the persistence of therapeutic effects following cessation of stimulation is driven by a dual-pathway mechanism: direct neural stimulation provides immediate sympathetic drive, while concomitant cortisol release outlasts the initial stimulus.

The present study employed 6 dpf larval zebrafish, a stage at which both cardiac and neural systems exhibit sufficient functional maturity for the intended optomechanical cardiac modulation^65,66. Moreover, the robustness of optomechano-cardiac response was further confirmed by control experiments conducted at 12 dpf (Supplementary Fig. 17), demonstrating that the fundamental auditory-cardiac neural pathway remains functionally conserved beyond early larval stage. Nevertheless, it must be acknowledged that cardiac innervation undergoes continuous maturation, with age-dependent changes in the total number and relative proportion of adrenergic neurons^67,68. Consequently, future studies comparing these responses across different developmental stages will be valuable to elucidate the maturation of this neural pathway and explore the broader applicability of the proposed opto-mechano-cardiac coupling strategy.

In conclusion, this work establishes a paradigm for opto-mechano-cardiac coupling to modulate cardiac rhythm, merging programmable optical manipulation with systems physiology to address three longstanding challenges: i) Disentangling overlapping auditory inputs through organ-specific stimulation; ii) Decoding music waveforms from an art into a quantitatively programmable biophysical stimulus; iii) Developing non-genetic cardiac neuromodulation towards personalized arrhythmia management. By organic integration of flexible programmability and excellent spatiotemporal precision unattainable through conventional acoustics, this proposed strategy advances music therapy from an empirical practice to a mechanism-driven discipline. Ultimately, this framework provides a potential alternative for mechanistic studies of sensory-cardiac integration and opens new avenues for personalized antiarrhythmic therapy.

Methods

Experiment setup

The experiment setup was constructed around a custom-modified inverted microscope (Nikon Eclipse Ti2-E). Firstly, the laser beam, at a wavelength of 1064 nm, was interacted with an acousto-optic deflector (AOD) to enable a desired spatiotemporal distribution with a maximum switching rate of 100 kHz. After that, the modulated laser beam was expanded by a beam expander to overfill the pupil of the microscope, reflected upward by the dichroic mirror, and then refocused into the sample chamber. The illumination light was emitted from a halogen light source, focused through a condenser and then irradiated onto the zebrafish larvae for a bright-field imaging. Crucially, two independent imaging paths were designed to allow concurrent otolith manipulation and heart rate observation from different angles. The optical tweezers for otolith oscillation were applied through a 60x water-immersion objective (CFI Apo 60X NIR, Nikon), which was connected to the primary high-speed camera (1T-01-PRIME-BSI-EXP, Teledyne Photometrics, America) via a tube lens for real-time imaging. Meanwhile, the heart rate was monitored laterally via a second optical path fitted with a 10x objective (CFI Plan Fluor 10X, Nikon) projecting onto a second scientific camera (MG-100, Mshot, Guangzhou, China) via a tube lens. This allowed us to simultaneously record the heartbeat from the side while the otolith was manipulated through the 60x objective.

Zebrafish care and treatment

Zebrafish larvae were purchased commercially from Nanjing Eze-Rinka Biotechnology Co., Ltd., and maintained in 100 ml Petri dishes under controlled experimental conditions (temperature: 28 °C ± 0.5 °C; photoperiod: 14 h light/10 h dark cycle). Specimens were fed with artificial diet three times daily. Aquarium water renewal was performed every 24 h through 50% volume replacement, with waste separation achieved using 100-µm nylon mesh filtration to remove excreta and residual feed. The Tg(cmlc2:EGFP; flk:mCherry) double-transgenic line was used to concurrently label the vasculature (red) and the heart (green) for identifying the cardiac chamber and its spatial relationship to the otolith. Meanwhile, the other transgenic strain of huc:GCaMP6f; nacre was used for visualization of the whole-brain calcium dynamics during optomechanical oscillation. The orientation of the embedded zebrafish was precisely modulated to accommodate different experimental requirements. For optomechanical cardiac modulation, larvae were oriented laterally to enable simultaneous visualization of both the heart and the otolith. For calcium imaging of brain activity, the fish were re-oriented to a dorsal recumbency position with the head tilted downward towards the objective, which is optimal for obtaining an en-face view of the brain for high-resolution fluorescence imaging on an inverted microscope. This was achieved by carefully manipulating the zebrafish within the low-melting-point agarose gel using a fine artist’s brush before the agarose solidified. This final orientation provided an unobstructed and en-face view of the brain for optimal fluorescence signal collection on the inverted microscope. During all experiments, larvae were maintained at 28.5 ± 0.5 °C using a microscope stage temperature control system (ThermoPlate, Tokai Hit, Japan) to ensure physiological stability. All experimental procedures strictly adhered to the ethical standards approved by the Experimental Animal Ethics Committee of Guangzhou University (Approval ID: [2024]050).

Optomechanical otolith oscillation

Two optical traps, generated by a 1064 nm laser beam, were positioned bilaterally adjacent to the target otolith using the AOD, which was programmed to scan the optical traps in a reciprocating manner along a defined trajectory, inducing oscillatory motion of the otolith through optical gradient forces. Key oscillation parameters, including displacement amplitude, oscillation frequency and angular direction, were precisely tuned in real time via a custom-written control interface. The otolith’s motion was simultaneously monitored using high-speed bright-field imaging at 1200 frames per second, enabling programmable adjustment of the stimulation parameters.

Oscillation frequency quantification

The optical traps were scanned at predefined frequencies by programming the AOD, and meanwhile, the high-speed image sequences of the otolith were captured for each scanning frequency. Based on that, the displacement values of otolith motion were obtained following the established methodology²³, where subpixel image registration by cross-correlation was used to extract otolith motion. After that, the otolith’s oscillation frequency was determined by using a custom MATLAB algorithm to track the otolith’s position in each frame and perform a Fast Fourier Transform (FFT) on the position-time data. The dominant frequency component from the FFT spectrum was identified as the oscillation frequency. At each scanning frequency, the mean oscillation frequency was calculated from 5 independent measurements with error bars indicating the standard deviation.

Musical encoding

The proposed optomechanical platform can drive otolith oscillations across a much broader spectrum (i.e., 0 ∼ 50,000 Hz). Leveraging this capability, the desired musical encoding has been performed depending on the equal-tempered scale (A4 = 440 Hz), i.e., the fundamental frequencies of Do (C4), Re (D4), Mi (E4), Fa (F4), Sol (G4), La (A4) and Ti (B4) were mapped with 261.6 Hz, 293.7 Hz, 329.6 Hz, 349.2 Hz, 392.0 Hz, 440.0 Hz and 493.9 Hz, respectively. Building upon this authentic frequency set, two auditory stimuli have been quantitatively defined by their distinct spectral composition and temporal dynamics: 1) the exciting music was characterized by broad spectral coverage and rapid transitions, i.e., the stimulus pattern jumps across the full set of frequencies (261.6 ∼ 493.9 Hz) with a fast rhythm of 60 beat per minute (bpm), meaning the music pattern changes once per second. 2). The soft music was characterized by restricted spectral range and smooth transitions, i.e., the stimulus pattern is confined to a narrower mid-frequency band (i.e., Mi ∼ Sol: 329.6∼392.0 Hz) with a slow rhythm of 20 bpm, meaning the music pattern changes only once every three seconds. Notably, the larval zebrafish exposed to soft and exciting music stimuli were from the same batch but were different individuals. This design was intentional to avoid potential carry-over or desensitization effects that could occur from the sequential application of two different auditory stimuli to the same animal. Using distinct individuals for each stimulus condition ensures that the measured cardiac responses are attributable specifically to the acoustic characteristics of each music type.

Heart rate analysis

This study employed a custom-developed edge detection algorithm for analyzing the heart rate of larval zebrafish, enabling automated cardiac boundary extraction, real-time cardiac motion tracking, and quantitative heart rate computation. First, the acquired heartbeat images underwent a Gaussian filtering to remove stochastic noise, followed by contrast enhancement between cardiac structures and background through histogram equalization or adaptive thresholding. Manual delineation of cardiac regions confines computation to defined regions of interest (ROIs). Subsequently, the edge detection algorithm computed pixel-wise gradient magnitudes (i.e., first-order derivatives), for which abrupt gradient transitions were identified as cardiac boundaries. Finally, cardiac phases were automatically classified based on geometric features of boundary coordinates, i.e., end-systole (minimal area) and end-diastole (maximal area), thus enabling autonomous derivation of quantitative heart rate statistics.

Calcium imaging

The transgenic zebrafish (huc: GCaMP6f; nacre) were immobilized by using 2% agarose, and their orientation was modulated to a dorsal recumbency position with the aim of obtaining an en-face view of the brain for high-resolution fluorescence imaging on an inverted microscope. For calcium imaging, the neural activity was recorded for the transgenic zebrafish that pan-neuronally expressed GCaMP6f (huc:GCaMP6f; nacre). The brain was illuminated using an LED source with a central wavelength of 473 nm through a 20 × objective (CFI Plan Fluor 20 × , Nikon). The emitted green fluorescence was collected by the same 20 × objective, passing through a dichroic mirror and being filtered by a 512–558 nm bandpass emission filter before being acquired by the primary camera. To resolve neural activity from the dorsoventrally separated VMN and TS/Th, an electrically motorized Z-stage was programmed to auto-alternately sweep between two predefined focal planes during continuous imaging, i.e., one plane encompassing TS/Th while the other plane encompassing VMN. Within each 50-ms imaging cycle, fluorescence signals from both focal planes were integrated into a single frame, yielding a sampling frame rate of 20 Hz. This design enables quasi-simultaneous sampling of both regions with minimal temporal offset.

Fluorescence image processing

The spatiotemporal propagation of neural activity was visualized using a frame-differentiation method to calcium imaging data. The onset of optomechanical stimulation was triggered at t = 2 s, which was set through a transistor-transistor logic (TTL) signal generated by the operation software. Following stimuli, a significant calcium fluorescence increase was observed starting at the frame captured at t = 2.6 s. Subsequently, the difference images were computed for five consecutive 0.2-second intervals (i.e., 2.6–2.8 s, 2.8–3.0 s, 3.0–3.2 s, 3.2–3.4 s and 3.4–3.6 s) by subtracting the grayscale values of the previous frame (F_t-1) from the current frame (F_t), highlighting pixels with newly increased fluorescence. Notably, the signal thresholding was introduced to eliminate subthreshold fluorescence variations, which should be attributed to spontaneous neuronal activity. The image sequences from two focal planes (i.e., the plane comprising VMN or TS/Th) were processed separately. The resulting difference images from both planes were spatially calibrated (based on anatomical landmarks) and integrated into a single composite map per time window. Finally, the five resulting difference images, each representing neural activity in a distinct temporal interval, were assigned to a discrete color (red, yellow, green, blue, purple) and spatially registered to a reference brain image using anatomical landmarks.

K-means clustering analysis

The used K-means clustering analysis was implemented using a Python-based computational pipeline with the following procedures: (1) Image registration: Motion artifacts induced by zebrafish spontaneous movements were corrected via phase-correlation-based rigid registration. Optimal translation parameters between each frame and a reference frame were computed to achieve whole-brain spatial alignment. (2) Clustering execution: Initial centroids were selected via K-means ++ initialization, followed by iterative assignment of data points to nearest centroids and centroid recalculation. Iterations terminated when centroidal displacement fell below 1-pixel tolerance. Ten independent clustering runs ensured robustness, with stability validated through consensus matrices. Neurons within identical clusters exhibited synchronous calcium dynamics, while distinct clusters showed divergent activity patterns. (3) Clustering analysis: Quantitative trajectory analysis of calcium transients was performed for three neuronal response regions. Resultant response data were visualized as time-series graphs, where the horizontal and vertical axes indicated the time and normalized calcium signal intensity, respectively. The response onset latency for each region was then determined as the time point at which the calcium signal showed a rapid increase, identified by locating the maximum slope of the rising phase. Individual delays from all fish were pooled and presented as mean ± SD in the bar graph, with each data point representing one fish.

Bradycardia model

A bradycardia disease model was established by co-incubating 6-dpf larval zebrafish in 50 μM doxorubicin-supplemented culture medium. Microscopic examination after 5-hour drug exposure revealed significantly depressed heart rates in treated larvae (1.5 Hz ± 0.05 Hz) compared to untreated controls (2.3 ± 0.05 Hz at rest), confirming successful recapitulation of pathological bradycardia.

AV block model

The AV block disease models were established by preparing a terfenadine stock solution followed by sonication for 10 min after adding aliquots into culture medium to achieve a final concentration of 5 μM. Subsequently, 5 dpf larval zebrafish were subjected to drug coincubation in the terfenadine-supplemented medium. After 24-hour exposure, larvae exhibiting successful atrioventricular (AV) block phenotypes were selected under microscopic screening. These model-validated larvae were mounted on the sample stage, where optomechanical otolith oscillation was performed using SOTs while cardiac dynamics were recorded pre- and post-oscillation.

Sinus pause model

The sinus pause model was constructed by co-incubating larval zebrafish in culture medium supplemented with 0.5 μM Sunitinib. Following a 5-hour drug exposure, microscopic examination revealed a transient sinus pause, i.e., indicative of escaped heartbeats. Drug concentration-dependent escalation of pause duration directly correlated with progressive impairment of cardiac functional reserve, where prolonged pauses signified advanced contractile dysfunction.

Pharmacological validation with isoproterenol

To test the role of β-adrenergic signaling in rescuing drug-induced arrhythmia, a systematic pharmacological intervention was conducted. The larval zebrafish exhibiting confirmed arrhythmias (doxorubicin-induced bradycardia, terfenadine-induced AV block or sunitinib-induced sinus pauses) were incubated in culture medium containing 10 μM isoproterenol for 100 min. Following incubation, the larvae were transferred to the translation stage for cardiac rhythm assessment. After the initial recording, the larvae were thoroughly washed and maintained in drug-free medium for a 100-minute washout phase. Post-washout cardiac function was re-evaluated using the same recording protocol.

Pharmacological inhibition of interrenal signaling via BAPTA microinjection

To investigate the potential hormonal contribution to the optomechano-cardiac response, a targeted microinjection approach was employed. Larval zebrafish were anesthetized and immobilized in low-melting-point agarose. A glass micropipette (Gaierdena Co., Ltd.; outer diameter: 1.14 mm, inner diameter: 0.5 mm) was pulled to a tip diameter of 0.5–5 μm using a micropipette puller and mounted on a programmable nanoliter injector (Nanojet III, Drummond Co., Ltd.). A stock solution of BAPTA-AM (a membrane-permeable calcium chelator) was prepared in dimethyl sulfoxide (DMSO) and subsequently diluted in physiological saline to a working concentration of 40 µM. A volume of 5–6 nL of this solution was precisely injected into the region of interrenal gland. This concentration was selected to achieve effective intracellular calcium chelation within the interrenal cells while minimizing systemic toxicity. A sham control group was injected with the same volume of physiological saline containing an equivalent concentration of DMSO. The injection process was monitored in real-time to ensure accurate targeting. Following injection, larvae were allowed to recover for 1 h prior to optomechanical stimulation assays to permit intracellular esterase conversion of BAPTA-AM to its active form.

Statistics and reproducibility

The statistical analysis was performed using GraphPad Prism (v8.0, GraphPad Software, Inc.). Results are presented as mean ± standard deviation (SD). The reported sample size refers to biological replicates (i.e., independent zebrafish embryos), ensuring conclusions reflect inter-individual variability. Comparison between two groups was performed by a two-tailed ttest with Welch correction, assuming a Gaussian distribution. Comparisons among multiple groups were performed by one-way ANOVA with Bonferroni correction. A P < 0.05 was considered to be statistically significant. Sample sizes were determined based on preliminary experiments, and the numbers used were also clearly reported in the relevant figure legends. In addition, no data were excluded from the analyses, and the zebrafish were randomly allocated into each treatment group. Due to the interdisciplinary aspect and the proof-of-concept development nature of this study, true blinding of experiments was not possible. However, data collection and analyses for some experiments were performed by separate individuals, and meanwhile, these collectors/analysts were not aware which samples corresponded to which experimental groups at the time of data collection and analysis.

Reporting summary

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

Author contributions

X.L. conceived the idea, curated the data and wrote the main manuscript. H.Q. and Y.L. performed the otolith oscillation and arrhythmia therapy measurements. G.Y., M.C., H.R., and Z.J. performed neural signaling pathways analysis and heart rate characterization. H.Q. and G.Y. wrote the material and structural characterization part of the manuscript. X.L., C.S., X.Z., and B.L. supervised the project, reviewed the manuscript, acquired funding and provided resources. All authors discussed the results and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Natural Science Foundation of China (62375105, 52373287, and 62135005), the Guangdong Basic and Applied Basic Research Foundation (2025B1515020099 and 2023A1515030225), the Open Fund of Guangdong Provincial Key Laboratory of Nanophotonic Manipulation (No. 202401) and Guangdong S&T Program (No. 2023B1212010008), the Science and Technology Program of Guangzhou (No. 2025A03J3138), the Guangzhou Program on Basic and Applied Basic Research (2025A04J6004), the Tertiary Education Scientific Research Project of Guangzhou Municipal Education Bureau (2024312275), and the Fundamental Research Funds for the Central Universities (21623212 and 21626728).

Data availability

All data supporting the conclusions of this study are available within the article or the Supplementary Materials, and the source data are provided as a Source Data file. 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 https://doi.org/10.1038/s41467-026-76695-x.