Liquid metal-based stretchable bioelectronic fiber for electrical stimulation and drug delivery in minimally invasive cardiac therapy Open Access

Cardiovascular disease, the leading cause of death worldwide, often accompanies severe emergency situations, which highlights the need for effective and urgent interventions to improve patient survival rate.^1,2 Current immediate care strategies, including external high-energy electrical shocks and the administration of pharmaceutical agents, have contributed to saving patients' lives,^3–5 but they often result in systemic side effects^6–8 due to their impact on the entire body. For instance, the electrical shock, used in defibrillation therapy, delivers high-voltage impulses (∼360 J) to the chest to reinitialize the cardiac rhythm. This high-intensity stimulation, while effective, can induce complications, such as skin burns, myocardial injury, and pro-arrhythmic consequences. The pharmacotherapy, usually administered orally or by injection, is also prone to systemic side effects, such as electrolyte imbalances, unintended impacts on the nervous system, liver toxicity, and hypotension.

To minimize side effects, implantable bioelectronics have been considered as a promising solution for localized and acute interventions.^9–23 These electronics can offer spatially selective modulation to the target region of the body with immediate therapeutic responses. Furthermore, it can provide closed-loop personalized electrical modulation for individual symptoms on the basis of continuous monitoring of cardiac electrograms. However, the inherent rigidity and bulky design of conventional bioelectronics often cause unintended tissue damage as well as performance deterioration from mechanical mismatch of the device and the tissue.

Recent advances in soft-material-based implantable bioelectronics have partially addressed these issues.^{10,11,17,21,23} Among various soft materials,^24–34 liquid metals and nanocomposites have emerged as promising candidates for soft bioelectronics due to their remarkable mechanical and electrical properties. Liquid metals exhibit exceptionally high conductivity (∼3.4 × 10⁶ S/m) that remains stable under repetitive mechanical deformations.^30,35–41 They are in a liquefied state at physiological temperatures, so their high softness enables seamless integration into soft curvilinear tissues. Specifically, gallium-based liquid metals are known for their biocompatibility, allowing safe interaction with biological tissues, making them ideal for implantable devices.^36,42 Additionally, nanocomposites, composed of elastic polymer and conductive fillers, provide high conductivity as well as stretchability.^{20,32,43–47} They also feature tunable functionalities through the incorporation of various nanoparticles, demonstrating low impedance, excellent mechanical durability, and even additional functionalization such as enhanced biosensing capabilities or improved biocompatibility. Despite these advantages, the complexities of soft-material-based bioelectronics still require more intricate and invasive surgical procedures, limiting their use to a small subset of patients. For broader applications such as emergent cardiac conditions, strategies enabling rapid and straightforward implantation are required.

Moreover, conventional cardiac implants rely solely on electrical therapy. Such devices cannot accommodate simultaneous pharmacological treatments due to challenges in drug refills and the need for additional space for drug reservoirs. However, cardiac drug administration plays a critical role not only in managing arrhythmia but also in non-arrhythmic cardiovascular disorders, supporting heart recovery during or after electrical modulation.^48,49 Thus, multimodal therapeutic approaches integrating electrical and pharmacological therapies within an all-in-one cardiac implantable device is essential for a more comprehensive and effective treatment.

Fiber-configured soft-material-based devices present a promising solution for integrating multiple treatment modalities while offering a desirable structure for minimally invasive implantation.^{15,17,50–54} Their thin, elongated structure supports multifunctionality, enabling both precise electrical modulation and drug delivery. Additionally, it facilitates easy connections between the target organ and the external electrical systems, minimizing damage at the insertion site. Although these devices have been widely applied to organs such as the brain or gastrointestinal tract,^{15,17,50,52,53} their application to the heart remains limited due to the challenges of accommodating the heart's dynamic movements. To design effective fiber-based cardiac implantable devices, three critical requirements must be addressed: (1) incorporation of multiple functions to manage diverse cardiac symptoms or conditions; (2) stable performance under repetitive mechanical deformation from dynamic movements of the heart; and (3) specific strategies for handling soft devices while ensuring minimally invasive implantation.

Here, we present a stretchable, multimodal fiber that can be administered via minimally invasive implantation onto the epicardium, capable of performing electrogram recording, electrical stimulation, and drug administration. Utilizing liquid metal and functionalized nanocomposite, the device manifests stable performance even under the heart's dynamic and harsh conditions. The phase transition property of liquid metal enables rapid and less-invasive implantation, simplifying the surgical procedure while ensuring precise placement. Upon detecting abnormal signals during electrogram recording, electrical stimulations, drug delivery, or a combination of both can be applied to provide targeted treatment, with therapeutic effects monitored in real time. This closed-loop therapeutic system is successfully demonstrated in an in vivo rat model.

The device mainly features a tri-layer structure comprising a central drug channel, an intermediate layer of liquid metal (eutectic gallium–indium, EGaIn), and an outer nanocomposite layer [Fig. 1(a-i)]. The nanocomposite, made of gold-coated silver nanowires (Ag–Au NWs), functions as a biocompatible and low impedance epicardial electrode, leveraging the inertness and high biocompatibility of gold, without losing intrinsic high conductivity of silver [Fig. 1(a-ii) and supplementary material Fig. 1]. Both the nanocomposite and EGaIn exhibit stretchability and strain-insensitive electrical properties, enabling the fiber to reliably transmit electrical signals even in dynamically deforming environments.

The innovative administration method of the fiber, minimally invasive implantation, could be implemented by exploiting the low melting point of EGaIn (∼15 °C). When cooled below 0 °C, the fiber becomes rigid, facilitating easy penetration through the skin without invasive surgical tools. Upon reaching body temperature, the EGaIn transitions to low-modulus state, ensuring soft and conformal contact with the delicate epicardial tissue, while minimizing the risk of damage [Fig. 1(a-iii)]. Meanwhile, the central microfluidic channel, fabricated from Ecoflex silicon with an elastic modulus similar to that of the nanocomposite, maintains structural integrity and stable performance under mechanical deformation, thereby preventing delamination. Additionally, the fiber is encapsulated with Ecoflex elastomer, leaving both ends exposed. This supplementary encapsulation layer serves to prevent unintentional interaction with surrounding tissue at the insertion site during the delivery of biophysiological or electrical signals.

By integrating microfluidic channels with conductive materials, the fiber achieves three key multimodal functions: (1) real-time electrogram recording, (2) electroceutical stimulation for immediate intervention, and (3) microfluidic drug delivery directly to the heart to regulate abnormal heart rates and complement electroceutical therapies [Fig. 1(b)]. Electrical stimulation, which provides an immediate effect on heart rate, is particularly effective in acute conditions such as heart failure, cardiac arrest, or ventricular arrhythmias. However, in several cardiovascular diseases, such as atrioventricular block or sick sinus syndrome, electrical stimulation by itself may be insufficient.⁵⁵ Pharmacological treatment becomes especially critical in such scenarios, as it could regulate abnormal heart rates that cannot be fully addressed by electrotherapy. Moreover, combining electrical and pharmacological therapies may improve therapeutic outcomes by alleviating complications such as thrombosis or pain associated with only electroceutical interventions.⁵⁶ 

Due to its multifunctionality and ease of implantation, the fiber is well-suited for emergency applications. When cardiac emergencies occur, the fiber can be rapidly implanted using the phase transition properties of the liquid metal. Once implanted, the all-in-one device provides treatment solutions tailored to the condition of patients by monitoring epicardial signals in real time, ensuring an acute and effective closed-loop therapeutic system [Fig. 1(c)].

The elastomeric nanocomposite is prepared through a multi-step process. First, Ag–Au NWs are synthesized by sheathing Ag NWs with gold as previously reported.^46,47 The Ag–Au NWs are then uniformly dispersed in a solution of toluene and ethanol with hexylamine as a dispersant. Separately, styrene–butadiene–styrene (SBS) block copolymer is dissolved in toluene to prepare an elastomer solution. These two solutions are mixed thoroughly via vortexing to form a nanocomposite solution. The mixture is subsequently mold-cast and dried, yielding a nanocomposite with a dense percolation network of Ag–Au NWs embedded within the SBS matrix. This configuration provides the material with excellent electrical conductivity, stretchability, and biocompatibility [Fig. 2(a-i)]. The phase-convertible liquid metal, EGaIn, is encased within the nanocomposite shell [Fig. 2(a-ii)].

The fabrication process of the fiber is illustrated from Figs. 2(b)–2(d). First, the microfluidic channel is prepared by mold-casting Ecoflex. A long metal wire is positioned at the center of a 3D-printed mold, Ecoflex is poured in and cured, and the wire is subsequently removed to create the hollow microfluidic channel [Fig. 2(b)]. Next, a polydimethylsiloxane (PDMS) mold is prepared to shape the fiber. The PDMS mold design, shown in Fig. 2(c) includes a central section to define the fiber and grooves at both ends for positioning the microfluidic channel centrally within the fiber's cross section.

The final fiber is fabricated by integrating the components described above [Fig. 2(d)]. The nanocomposite solution is blade-coated onto the PDMS mold and dried under ambient conditions, forming a U-shaped nanocomposite film due to capillary forces between the solution and mold surface. EGaIn is then injected into the voids enclosed by the nanocomposite film. The pre-fabricated microfluidic channel is placed along the grooves in the mold, aligning it at the center of the EGaIn layer. A second nanocomposite layer is blade-coated over this structure, encapsulating the EGaIn and microfluidic channel within the nanocomposite shell. After sufficient coating, the edges of the mold are removed, and both ends of the device are additionally sealed with the nanocomposite material to prevent any leakage of the liquid metal. Finally, to prevent unintentional electrical stimulations in the surrounding tissues, the main shaft is insulated by casting Ecoflex elastomer.

Von Mises stress analysis highlights the penetration capabilities of the fiber in its solidified and liquefied states [Fig. 3(a)]. In the solidified state, the fiber exhibits a high elastic modulus (500 MPa) that is superior to the modulus of the breast tissue (3.43 MPa)^57–59 enabling effective penetration on skin for fiber's minimally invasive characteristics. Furthermore, the fiber could maintain structural integrity even under frozen process, as the induced stress remains below the fracture strength (1.66 MPa). Conversely, in the liquid state, the elastic modulus drops to 5 MPa, which implies that the insertion of fiber in liquefied state is impossible. The maximum principal stretch represents the specific ratio of the stretched length to the original length for both solidified and liquefied fiber (supplementary material Fig. 2). These mechanical properties are experimentally validated through tensile tests [Figs. 3(b) and 3(c)], revealing a maximum strain of 250% in the liquid state, indicative of high ductility, and a strain tolerance of up to 50% in the frozen state, reflecting brittle behavior.

The fiber exhibits high stretchability, a critical property for cardiac bioelectronics, with elongation up to 300% without any defects [Fig. 3(d)]. Given the natural stretchability of the heart (∼20%),²⁴ the stretchability of the fiber is sufficient to withstand the dynamic motion on the heart. In addition, such external deformation rarely affects the electrical property of the fiber. Strain-insensitive electrical properties, essential for maintaining stable performance under extreme conditions, are confirmed in Figs. 3(e)–3(g). The relative resistance change remained below 1 under 80% strain, indicating minimal variation in the resistance under mechanical deformation, making the fiber well-suited for cardiac applications. Additionally, cyclic tests under 30% strain demonstrate consistent electrical performance over repeated loading cycles, validating the durability and reliability of the fiber.

The device's capabilities as bioelectronics are further verified through electrochemical tests [Fig. 3(h)]. Impedance measurements over a 5-h period reveal minimal changes at 1000 Hz, a commonly used frequency for biophysiological recording, demonstrating electrochemical stability [Fig. 3(i)]. Similarly, periodic current density changes in response to cyclic voltage confirm the fiber's durability and repeatability in dynamic environments, underscoring its suitability as a capacitive electrode [Fig. 3(j)]. Cyclic voltammetry analysis reveals a nonlinear current–voltage relationship, showcasing the fiber's combined charge storage and transport properties, further highlighting its functionality as a capacitive electrode [Fig. 3(k)].

The fiber delivers drugs to the epicardial surface through its hollow microfluidic channels, with a diameter of 400 μm. Utilizing a straightforward fabrication process, the fiber was designed with multiple drug channels, enabling the simultaneous administration of different drugs (supplementary material Fig. 3). Before use, the fiber should be solidified with a freezing spray to enable skin penetration. After implantation, it re-softens, allowing seamless contact with the epicardial tissue without any damage [Fig. 4(a)]. Since the fiber starts to melt after approximately 40 s of tissue contact, there is sufficient time to reach the epicardium (supplementary material Fig. 4). An ex vivo demonstration of the sequential drug delivery process is shown in Fig. 4(b), by using 1 wt% agar hydrogel as an artificial tissue model. The fiber, equipped with a sharp tip and enhanced stiffness due to solidified EGaIn [Fig. 4(c)], could penetrate the artificial tissue to a depth of 3 cm, which is enough to be clinically applicable. Once implanted, it can be softened and slightly bent, while continuing to facilitate the sequential delivery of two mock drugs.

To validate its suitability for in vivo application, we evaluated the fiber's drug delivery performance under various conditions [Fig. 4(d)]. The 3D-printed mold allows the fiber to be connected to commercial syringes of various capacities (supplementary material Fig. 5). Among these, a 1 ml syringe was used for flow rate measurements conducted under different conditions. First, the flow rate of the mock drug through the channel was measured under different injection pressures. The flow rate of the fiber was comparable to those of a conventional 31G syringe, ensuring precise dosing control on the epicardium [Fig. 4(e)]. Next, the fiber's performance was tested under bending conditions, maintaining a consistent flow rate even at a bending radius as small as 0.5 cm [Fig. 4(f)]. Additionally, the fiber exhibited reliable drug release performance under repetitive stretching [Fig. 4(g)]. These ex vivo experimental findings demonstrate the fiber's suitability for cardiac applications, particularly in dynamic physiological and constrained environments. It ensures precise dosing and maintains durability and functionality during implantation, making it an effective tool for epicardial drug delivery.

To validate the in vivo feasibility of the fiber for cardiac monitoring and electrotherapy, we demonstrated its minimally invasive implantation, electrophysiological monitoring, and therapeutic application in a cardiopathologic animal model [Fig. 5(a)]. The sterilized fiber was frozen using a simple ice spray, and the sharpened tip was gently inserted into the thoracic cavity of Sprague–Dawley rats through the intercostal space between the left 3rd and 4th ribs. The implantation process was captured via endoscopic imaging [Figs. 5(b)–5(d)].

The epicardial electrogram (hereafter referred to as “epicardiogram”) was recorded and compared to the surface electrocardiogram (ECG) on the same scale to assess recording quality [Fig. 5(e)]. Unlike the conventional surface ECG (black), the epicardiogram (blue) provided clearer signals with higher amplitude (30 mV compared to 0.7 mV) and a superior signal-to-noise ratio (SNR 122 dB compared to SNR 14 dB). This suggests that the epicardiogram can offer spatiotemporally precise information on local cardiac conduction that is unattainable with conventional surface ECGs. Further implementation of an epicardiogram recording in an in vivo rabbit model validates the fiber's capabilities as an electrode even in larger animals (supplementary material Fig. 6). Additionally, cardiopathologic conditions, such as ventricular arrhythmias, were detectable through the epicardiogram. For example, ventricular bradycardia was induced by intraperitoneally injecting diltiazem (100 mcg per kg). Before injection, the epicardiogram displayed a normal sinus rhythm with an R–R interval of 179.8 ± 8.4 ms. After injection, the R–R interval increased significantly to 948 ± 14.7 ms, indicating severe bradycardia [Fig. 5(f)]. To demonstrate therapeutic efficacy, electrical stimulation (1.2 V amplitude, 5 ms pulse duration, 360 beats per minute) was applied through the fiber (red upward triangles). The ventricular contractions were successfully synchronized with the electrical pacing, restoring the R–R interval to 166.6 ms, within the normal range for rats.

The fiber also effectively addressed ventricular tachycardia, a potentially fatal arrhythmia [Fig. 5(g)]. Tachycardia was induced by injecting norepinephrine (10 mcg per kg) following left anterior descending coronary artery ligation. The fiber accurately recorded the disorganized, rapid contractions associated with tachycardia. Since the arrhythmia did not resolve spontaneously, overdrive pacing (2 V amplitude, 1 ms pulse duration, and 10 Hz frequency) was delivered via the fiber for 10-s intervals until cardiac resynchronization was achieved.

To extend its functionality beyond isolated electrotherapy, the fiber's drug delivery capabilities were evaluated. Its dual-channel microfluidic system allowed for sequential drug delivery onto the epicardial surface. As a demonstration of its multi-drug delivery capability, mock drugs (red and blue dyes) were injected sequentially and simultaneously [Figs. 6(a)–6(d)]. The first mock drug (red dye) was successfully delivered [Fig. 6(b)], followed by the second (blue dye) through a separate channel [Fig. 6(c)]. Simultaneous injection of both dyes produced a violet-like color [Fig. 6(d)], demonstrating the potential for hybrid drug delivery.

To test pharmaceutical applications, drugs with physiological effects were delivered to the epicardial surface. The fiber was implanted, ensuring the drug channels were in contact with the epicardial surface. At rest, the initial R–R interval was 281.96 ± 2.03 ms. Epinephrine (1 mcg per kg) was injected, increasing heart rate and cardiac pressure, reducing the R–R interval to 203.04 ± 0.18 ms. Subsequently, diltiazem (10 mcg per kg) was administered through the second channel, slowing the heartbeat and increasing the R–R interval to 417.32 ± 1.23 ms. To demonstrate the synergistic effect of electrotherapy and chemotherapy, electrical stimulation was applied to restore the heartbeat to normal ranges, reducing the R–R interval to 166.7 ms. All R–R interval changes were statistically significant (p < 0.005) [Figs. 6(e) and 6(f)].

To evaluate the physiological feasibility of epicardial drug delivery, diltiazem was administered to seven groups of mice (n = 5 per group) via different methods: subcutaneous (s.q.), intramuscular (i.m.), intraperitoneal (i.p.), intravenous (i.v.), and epicardial injection (a dose of 10 mcg/kg was administrated for s.q., i.m., i.p., and i.v., and three levels of dose—10, 5, and 1 mcg/kg—were used for epicardial injection) [Figs. 6(g)–6(i)]. Among conventional methods, intravenous injection exhibited the fastest effect. Interestingly, epicardial injection provided an even faster response than i.v. administration, even at lower doses (10%). These results highlight the fiber's potential for precise, localized drug delivery with reduced dosage requirements, making it a promising tool for addressing cardiac emergencies.

For clinical translation, however, additional advancement is required. In this study, as clinicians evaluated the disease based on recorded electrogram and implemented treatments, the time delay was inevitable, limiting the device's potential application in emergent situations. By incorporating automatic disease classification and treatment using an add-on device, significantly faster and more precise interventions could be achieved. To demonstrate this potential, we developed a cardiac disease classification code using MATLAB with the epicardial electrogram signals collected from the fiber, as shown in Fig. 6(j) and supplementary material Fig. 7. This MATLAB code processes the signal to enhance the QRS complex and detects actual R-peak positions within segments that exceed a predefined threshold. The heart rate is then calculated based on the number of detected R-peaks and the total duration of the recording. By comparing the calculated heart rate with known reference values for normal rat heart rates, cardiac disease classification can be effectively performed.

In this study, we developed a stretchable fiber optimized for cardiac emergency applications, integrating multimodal functionality with a design tailored for rapid and straightforward implantation. Constructed from a nanocomposite and EGaIn, the fiber demonstrates exceptional stretchability and strain-insensitive electrical properties, enabling continuous electrogram recording and reliable electrical signal delivery, even under dynamic motion of the heart. The fiber's adaptive modulus, achieved through the temperature-responsive phase transition behavior of EGaIn, facilitates minimally invasive and instantaneous implantation. Furthermore, the incorporation of a microfluidic channel into the fiber allows for stable and durable delivery of multi-drugs, enhancing potential treatment efficacy. Throughout the in vivo experiment, the fiber enabled continuous epicardiogram monitoring and targeted treatments based on arrhythmia manifestations, effectively demonstrating the efficacy of a close-loop therapeutic system. Notably, the faster response time of direct epicardial drug administration compared to conventional injection methods highlights the potential as a rapid and efficient intervention for cardiac emergencies.

Solvents such as ethylene glycol, ethyl alcohol, acetone, and toluene were purchased from Samchun Chemicals (Republic of Korea). Poly(styrene-butylene-styrene) (SBS, KTR 103) was acquired from Kumho Chemicals (Korea). Liquid metal (eutectic gallium–indium) and hexylamine were purchased from Sigma Aldrich (USA). Ecoflex elastomer was obtained from Hyup-shin corporation. Gold-coated silver nanowires were synthesized as previously reported.^43,47 The connector between the conventional syringe and the fiber was polyJet 3D-printed using VeroWhite photopolymer resin.

The nanocomposite was prepared by mixing Ag–Au nanowires with SBS polymer. The SBS polymer was dissolved in toluene solvent at a weight ratio of 1:10 to form a polymer solution. Simultaneously, a solution of Ag–Au nanowires was prepared at a concentration of 50 mg/ml in a mixed solvent of ethanol and toluene at a ratio of 1:3. The two solutions were combined and thoroughly mixed using vortexing, followed by the addition of hexylamine (20 μl) as a dispersant.

The tissue was modeled as a cubic box with a side length of 70 mm. The base of the cube was determined as a fixed surface, meaning it could not undergo deformation. All other surfaces, except for the fixed surface, were allowed to deform under loading. The fiber was modeled as a rectangular box with dimensions of 2 mm in the x-direction, 2 mm in the y-direction, and 100 mm in the z-direction and the tip was beveled at an angle of 20°. The base of the fiber was in contact with the top surface of the tissue, with their centers aligned at the same point. Young's modulus was set to 25 kPa for the tissue, 5 MPa for the fiber in the liquefied state, and 500 MPa for the fiber in the frozen state, as determined from UTM measurement. Poisson's ratios of 0.5 for the tissue and 0.43 for the fiber, in both solidified and liquefied states, were applied. A force was loaded on a free xy-surface of the fiber in the z-direction until the surface moved by 5 mm in the z-direction.

A tensile test was conducted to evaluate the stretchability and compare the elastic modulus of the fibers in dual state; solidified and liquefied. The test was performed using a UTM with the stretching speed of 50 mm/min, following a modified ISO 527. The resistance of the fiber under strain and during cyclic experiments was measured using a four-point probe connected to a source meter (Keithley 2450). For sample preparation, the fiber was transferred onto the VHB tape, and Cu wires were attached to the fiber's surface using silver paste for electrical connection. In the strain test, the fiber was initially 1 cm long and stretched by 1 mm increments up to 100% strain for ten cycles. In the cyclic test, the fiber with an initial length of 1 cm, was subjected to 30% strain for 1000 cycles at a frequency of 0.1 Hz.

Electrochemical properties of the fiber as a bioelectronic were evaluated using an electrochemical workstation (CHI-660E, CH instruments). The fiber was immersed in a PBS solution with an exposed area of 45 mm². Two Pt electrodes were used as reference and counter electrodes, respectively. Impedance measurements were performed over a frequency range of 1–0.1 MHz with an amplitude of 5 mV. The charge injection capacity was determined by applying an AC pulse with an amplitude of 0.5 V, a pulse width of 0.2 s, and a sensitivity of 0.01 A/V. Cyclic voltammetry curves were obtained from −0.5 to 0.5 V at a scan rate of 100 mV/s.

All animal experiments were performed under approval of Seoul National University IACUC (SNU-221102-3). The 8-week-old male SD rats (Orient Bio Inc., Korea) were anesthetized with 3%–5% isoflurane chamber. Hair removal using clipper was followed by administration of the commercialized hair removal cream on the left thoracic region. The skin of the intercostal region between the 3rd and 4th rib was poked with an 18G needle. After two microfibers were implanted on the desired location of the epicardial surface, other ends of the microfibers were connected to the commercialized DAQ device, Power Lab (AD Instruments, New Zealand). Norepinephrine was injected intravenously to induce tachyarrhythmias. Otherwise, the diltiazem solution was injected intravenously to induce bradyarrhythmia. To deliver electric stimulation, the pair of microfibers was connected to the current output channel of the DAQ device, Power Lab (AD Instruments, New Zealand). The surface ECG was recorded during stimulation.

All data are analyzed by the two-sample t test. All statistical indication in the graph presents as mean with standard deviation. Sample size (n) = 5 for experimental data, unless otherwise noted in figure caption. Statistical details such as p value are described in each figure caption.

See the supplementary material for additional information.

This work was supported by the Institute for Basic Science (No. IBS-R006-A1); the Nano & Material Technology Development Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Science and ICT (No. RS-2024-00450561); and the NAVER Digital Bio Innovation Research Fund, funded by NAVER Corporation (Grant No. 37-2023-0040).

The authors have no conflicts to disclose.

Ethics approval for experiments reported in the submitted manuscript on animal or human subjects was granted. All animal experiments were performed under approval of Seoul National University IACUC (SNU-221102-3).

Chanhui Park, Yesol Kim, and Seonghyeon Nam contributed equally to this work.

Chanhui Park: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Yesol Kim: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (lead); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Seonghyeon Nam: Conceptualization (equal); Data curation (equal); Investigation (supporting); Methodology (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Hyejeong Kang: Validation (supporting). Joonho Moon: Validation (supporting); Writing – review & editing (equal). Ji Hoon Kim: Validation (equal); Visualization (supporting). Gi doo Cha: Validation (supporting); Writing – review & editing (equal). Seung-Pyo Lee: Funding acquisition (lead); Methodology (equal); Project administration (equal); Writing – review & editing (supporting). Sung-Hyuk Sunwoo: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Project administration (equal); Supervision (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Dae-Hyeong Kim: Funding acquisition (equal); Project administration (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.