A sonoelectric artificial meningeal device for enhanced implantable electric field ablation of glioblastoma

Abstract

Electric field tumor therapy has emerged as a promising non-invasive treatment for glioblastoma (GBM), but its clinical efficacy is severely limited by skull-induced attenuation of electric field intensity. To overcome this limitation, we developed a multilayer thin-film sonoelectric meningeal device (MF-SMD) consisting of an electrospun PLGA artificial meningeal scaffold integrated with a microneedle-based triboelectric film. This device converts externally applied low-intensity ultrasound into intracranial electric fields, enabling transcranial, localized, and controllable electric ablation of brain tumors. Upon low-intensity ultrasound excitation, the film generates localized intracranial electric fields that disrupt tumor mitosis while preserving neuronal integrity, and significantly reduce skull-induced field attenuation. In vitro and in vivo experiments validated the antitumor efficacy of the device, achieving approximately 68% proliferation inhibition in clinical GBM cell lines and a 72% reduction in tumor burden in orthotopic mouse models. Notably, compared with conventional tumor treating fields (TTF) systems that employ extracranial electrodes, the MF-SMD maintained electric field attenuation below 20% at therapeutic frequencies. These results establish a novel therapeutic paradigm that overcomes skull-induced electric field attenuation, addressing a key challenge in transcranial brain tumor therapy.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-58372-7.

Introduction

Glioblastoma (GBM) is a highly aggressive malignant brain tumor; despite the widespread use of surgery, radiotherapy, and chemotherapy, patient survival remains difficult to substantially improve^1–9. In recent years, tumor treating fields (TTFs) have emerged as a noninvasive modality that suppresses tumor growth and migration by disrupting mitosis through alternating electric fields, offering a promising adjunct for GBM therapy^10–13. However, the conventional TTFs faces key technical barriers: pronounced skull-induced attenuation and current shunting through the highly conductive scalp layer collectively reduces the electric field that reaches intracranial lesions, weakening therapeutic efficacy^14–18. In addition, external electrodes and percutaneous wiring complicate electrode positioning, hinder daily use, and increase patient discomfort and infection risk^19,20.

To overcome these limitations, wirelessly powered implantable devices have attracted increasing attention^21,22. Among them, ultrasound-based wireless energy transfer is particularly suitable for intracranial bioelectronic systems due to its deep penetration, strong spatial focusing, and favorable biosafety^23–28. However, existing ultrasound-based solutions face critical limitations: most utilize rigid or planar substrates that lack mechanical compliance with the soft brain surface, leading to chronic inflammation and foreign body responses; they still rely on transcranial electric field delivery without explicitly compensating for skull-induced attenuation; planar transducer designs suffer from limited energy conversion efficiency, often requiring higher ultrasound intensities that raise safety concerns; and many employ non-degradable materials, necessitating secondary surgical removal^18,28–30. By converting ultrasonic energy into localized intracranial electric fields, this approach enables precise and sustained treatment of brain tumors while avoiding skull attenuation inherent to conventional TTFs. It is important to note that the mechanism of our device is fundamentally different from direct high-intensity focused ultrasound (HIFU) ablation or focused ultrasound stimulation. HIFU ablation relies on thermal coagulation to destroy tumor tissue, whereas our device converts low-intensity ultrasound into alternating electric fields that specifically disrupt tumor cell mitosis—a non-thermal, anti-mitotic mechanism clinically validated in tumor treating fields (TTFields) therapy. Importantly, the added value of our approach is not about reducing ultrasound attenuation—which is already low in rodents—but about enabling intracranial TTFields therapy, which focused ultrasound alone cannot achieve.

Here, we propose and construct a multilayer thin-film sonoelectric meningeal device (MF-SMD) that can be placed into the surgical bed immediately following tumor resection, leveraging the existing surgical window without requiring an additional procedure. The device employs an electrospun PLGA-based artificial meningeal scaffold integrated with a multilayer triboelectric nanogenerator (TENG) film that converts low-intensity transcranial ultrasound into stable alternating electric fields, enabling localized and controllable therapy at the tumor site. The MF-SMD exhibits excellent biocompatibility and mechanical compliance for conformal, long-term implantation on the brain surface, and demonstrates significant anti-proliferative effects in both in vitro studies and orthotopic glioma mouse models. Moreover, the MF-SMD bypasses skull-induced attenuation by directly generating electric fields intracranially, while reducing complications associated with traditional electrode-based therapy (Fig. 1).

Fig. 1.

Schematic illustration of the cancer therapy with MF-SMD. (a) Comparison of electric field attenuation when the device is placed outside versus inside the skull. (b) Schematic of the MF-SMD. The device is designed as an implantable biomedical device that is excited by an external ultrasound power source and produced adjustable alternating electric fields to inhibit tumor growth in the brain. (c) Working mechanism of MF-SMD. (d) Biological mechanism of MF-SMD. All schematic illustrations in this manuscript were created by the authors using Microsoft PowerPoint and Adobe Photoshop.

Results

Design and working principle of the multilayer thin-film sonoelectric meningeal device (MF-SMD)

In tumor-treating field (TTF) therapy, the shielding effect of the skull remains a major obstacle to therapeutic efficacy. As illustrated in Fig. 1a, when the device is positioned extracranially, the electric field undergoes substantial attenuation while traversing the skull, leaving insufficient field strength within deep brain tissues to exert meaningful antitumor effects. By contrast, when placed intracranially, the field is no longer hindered by skull attenuation, allowing effective penetration into brain tissue, thereby enhancing tumor inhibition and inducing apoptosis. Building on this concept, we propose a Multilayer Film Sonoelectric Meningeal Device (MF-SMD) that not only functions as a subcranially implanted artificial meningeal scaffold that enables the generated therapeutic electric fields to bypass the skull completely, avoiding the dramatic attenuation inherent to transcranial electric field delivery, but also generates therapeutic electric fields upon activation by low-intensity external ultrasound (Fig. 1b). These electric fields disrupt the biological processes underlying tumor cell division, thereby enabling localized and controllable treatment of brain tumors^31–34. Ultrasonic excitation induces periodic contact-separation motion between the microneedle array and the counter electrode within the multilayer architecture, as supported by our finite element simulations (Fig. @@4e showing ~ 150 µm displacement under 0.2 W/cm² excitation). This mechanical motion results in the generation of a tunable alternating electric field with an intensity of 1–3 V/cm and a frequency of 100–300 kHz. The core working mechanism of the MF-SMD is illustrated in Fig. 1c. Under ultrasound excitation, periodic contact-separation occurs between the internal electrodes and the triboelectric layer. This process induces dynamic redistribution of interfacial charges, thereby generating alternating electric fields. Field strength is modulated by changes in contact area: separation enhances the field, whereas contact diminishes it, giving rise to a vibration-regulated and structurally dynamic field-generation mechanism^35–37. As shown in Fig. 1d, the resulting TTFs selectively target tumor cell mitosis by acting on polarizable proteins and charged biomolecules that are essential for spindle assembly. This interference disrupts proper chromosome alignment and segregation, ultimately leading to cell-cycle arrest and apoptosis. Importantly, this therapeutic strategy exhibits proliferation-dependent selectivity—it primarily affects rapidly dividing tumor cells while sparing post-mitotic normal neurons (which do not undergo mitosis). This is not “targeted therapy” in the sense of ligand-receptor specificity (e.g., monoclonal antibodies), but rather a physical mechanism that exploits the differential mitotic activity between tumor and normal cells. The safety of this approach is supported by our experimental data showing no reactive astrogliosis or neuronal loss in healthy brain tissue following MF-SMD treatment (Fig. @@6c, Fig. S2).^10,11.

Structural design and in vivo sonoelectric performance of the MF-SMD

The core architecture features a biodegradable artificial meningeal scaffold fabricated from poly (lactic-co-glycolic acid) (PLGA), a well-established bioresorbable polymer with documented degradation profiles in biomedical implants^38,39. The scaffold was engineered to a thickness of approximately 300 µm to mimic native meningeal biomechanics while ensuring biocompatibility and gradual resorption. While the long-term degradation kinetics were not the primary focus of this study—given that PLGA degradation is a well-characterized process—the selection of PLGA aligns with the design goal of eliminating secondary surgical removal. Integrally embedded within this scaffold is a 4-layer sonoelectric conversion film (Fig. 2a), comprising a wireless ultrasonic energy converter, a separator layer, a current collector, and an insulating layer. The term “sonoelectric conversion film” refers to this entire multilayer stack, not merely the current collector. Specifically, this film comprises: (1) a PDMS polymer microneedle film serving as the wireless ultrasonic energy converter (primary triboelectric layer); (2) a separator layer (gap layer, ~ 100 µm) that defines the initial separation distance between the microneedle film and the bottom electrode; (3) a current collector (flexible electrode sheet) that collects and outputs the generated charge; and (4) an insulating layer that encapsulates the electrical components and separates them from the electrospun meningeal scaffold. Among them, a PDMS polymer microneedle film is used as the wireless ultrasonic energy converter, and a flexible electrode sheet serves as the current collector; the gap between the two is controlled by the separator layer, while the electrospun meningeal scaffold and the sonoelectric conversion film are separated by the insulating layer. This multi-layered design enables efficient transduction of external low-intensity ultrasound into localized electric fields, with the PLGA scaffold simultaneously serving as a protective barrier, mechanical stabilizer, and biodegradable carrier that eliminates the need for secondary surgical removal.

Fig. 2.

Design and working principle of the multilayer thin-film sonoelectric artificial meningeal system. (a) Schematic illustration of the system composition: biodegradable PLGA-based artificial meningeal scaffold (< 1 mm thickness) with integrated 4-layer sonoelectric conversion film. (b) Fabrication of a PDMS microneedle layer via a microstructured replica-molding process, SEM image scale = 500 µm. (c) Fabrication of a PLGA artificial meningeal scaffold layer via an electrospinning process. (d) Schematic of MF-SMD implantation on the brain surface of a glioma-bearing mouse model. (e) Schematic of in vivo electrical output and tumor-suppression effects under ultrasound activation.

To achieve efficient ultrasound-to-electric energy conversion, we engineered a functional layer featuring a precisely patterned conical microneedle array (Fig. 2b). This key component was fabricated using a high-precision microstructured molding technique, wherein a master template with inverse microneedle topography was first created via microfabrication, followed by replication onto a biocompatible polymer matrix through thermal curing and demolding. The resulting array exhibited uniformly distributed conical microneedles with optimized dimensions—typically 400 µm in height and 320 µm in base diameter—arranged with ~ 900 µm pitch between needles, as verified by scanning electron microscopy (SEM) imaging (Fig. 2b). The complete MF-SMD device consists of the following layers (from bottom to top): the PLGA electrospun meningeal scaffold (~ 300 µm), an insulating layer (~ 50 µm), a current collector (flexible electrode, ~ 100 µm), a separator layer (~ 100 µm), and the PDMS microneedle layer (~ 400 µm). The total thickness of the device is approximately 950 µm (< 1 mm). This geometric design leverages the stress concentration effect of tapered structures: under ultrasonic excitation (28–200 kHz frequency), the microneedle tips undergo localized mechanical deformation amplitudes higher than planar films. This enhanced deformation induces robust shear stress at the interface between the triboelectric layer (PDMS microneedle film) and the conductive backing (current collector), promoting efficient separation of bound charges and generating a surface electric field output with peak intensity increased by 4 times compared to flat control samples (Fig. S1).

To address the critical challenges of tissue conformity and long-term implantation stability, we further developed an electrospun artificial meningeal scaffold using poly (lactic-co-glycolic acid) (PLGA) as the base material^40–45. The fabrication process employed a high-voltage electrospinning technique with a PLGA solution, yielding a nanofibrous membrane with a three-dimensional porous architecture (Fig. 2c). Scanning electron microscopy (SEM) characterization revealed directionally aligned nanofibers with an average diameter of 0.347 µm and interconnected pores, resulting in a structure that mimics the extracellular matrix (ECM) of native meninges. This nanofibrous network imparts dura-like mechanical properties, with a Young’s modulus of ~ 75 MPa and ultimate tensile strength of 4.7 MPa, closely matching the softness and flexibility of human dura mater (native range: 70 ± 44 MPa, tensile strength of 7 ± 4 MPa)⁴⁶. Functionally, the scaffold serves as a biomimetic interface between the implanted device and brain tissue: the high porosity facilitates nutrient diffusion and waste exchange, while the nanofibrous topography promotes the adhesion and infiltration of fibroblasts and endothelial cells, accelerating neovascularization and integration with surrounding neural tissue^47–49. This design synergistically combines mechanical support for the integrated sonoelectric conversion film with biological functionality, enabling seamless device-tissue interaction while promoting physiological healing. The MF-SMD device—comprising the electrospun PLGA scaffold integrated with the conical microneedle array layer—was affixed to the skull surface directly over the tumor region using biocompatible cyanoacrylate adhesive (Fig. 2d). A cranial window (1.5 mm diameter) was created to ensure acoustic coupling, with ultrasound gel applied between the transducer and device surface. As shown in Fig. 2e, under ultrasonic excitation, the device generates an alternating electric field suitable for electric field therapy of glioblastoma, which can induce tumor cell apoptosis without damaging normal nerve cells.

Electrical performance of MF-SMD

To verify the mechanical matching of the device substrate with brain tissue and to quantify its sonoelectric output, we first characterized the morphology and mechanics of the electrospun PLGA meningeal membrane. SEM revealed a continuous, uniform nanofiber network with high porosity, and the nanofibers exhibit higher orientation under high-speed collector conditions (Fig. 3a); fiber diameters followed an approximately normal distribution with an average of 0.347 ± 0.122 µm (Fig. 3b), favoring conformal tissue adherence. Uniaxial tensile testing (Fig. 3c) showed that increasing the collector rotation speed from 280 to 2800 rpm raised the ultimate tensile strength from 3.36 to 4.70 MPa, while the strain at maximum load decreased from 20.44 to 13.47%, and the fracture strain decreased from 23.03 to 16.87%, indicating that higher fiber alignment enhances strength at the expense of extensibility. Using an alternating electrospinning strategy (switching between 280/2800 rpm) to build an orientation gradient, the film obtained with one alternation exhibited an ultimate strength of 1.94 MPa, a strain at maximum load of 14.57%, and a fracture strain of 17.09%; with two alternations, these values were 1.26 MPa/7.50%/25.44%, respectively. Compared with previously reported uniaxial tensile data for adult human dura mater (ultimate strength ≈ 6.4 ± 1.8 MPa; maximum/rupture strain ≈ 11 ± 2%/14 ± 3%), the alternating-electrospun samples combine moderate strength with superior flexibility; notably, the twice-alternated specimen reached a fracture strain of 25.44%, demonstrating favorable compliance-buffering potential suitable for meningeal soft-tissue implantation^46,50.

Fig. 3.

Comprehensive evaluation of the mechanical and electrical of the MF-SMD. (a) SEM morphology of the electrospun PLGA meningeal membrane (Left panel: low-speed collector condition at 280 rpm; Right panel: high-speed collector condition at 2800 rpm; SEM image scale = 5 µm). (b) Fiber diameter distribution histogram (approximately normal); mean diameter 0.347 ± 0.122 µm. (c) Tensile stress–strain curves of electrospun PLGA membranes: i—collector speed 280 rpm (randomly oriented fibers); ii—collector speed 2800 rpm (highly aligned fibers); iii—alternating electrospinning once (280 ↔ 2800 rpm); iv—alternating electrospinning twice (280 ↔ 2800 ↔ 280 ↔ 2800 rpm). (d) Outputs of the MF-SMD under ultrasonic excitation of 0.4 W/cm² at a distance of 5 cm. (e) Field intensity variations with increasing distance from the ultrasound probe. (f) Tuning of MF-SMD output intensity by varying the ultrasonic excitation power.

The therapeutic efficacy of the MF-SMD hinges primarily on electric-field intensity, a parameter strictly regulated by clinical safety consensus to balance tumor inhibition and neural tissue protection. Based on decades of preclinical and clinical data—including landmark studies on tumor treating fields (TTFs) in glioblastoma—an intensity range of 1–3 V/cm is established as the “therapeutic window”^11,51,52: this range disrupts mitotic spindle assembly in rapidly dividing tumor cells while avoiding depolarization of resting neurons or damage to glial cells. Critical to clinical translation, our device achieves stable electric-field output within this window under low-intensity ultrasound drive, eliminating the need for bulky onboard circuits or size escalation. Instead, intensity tuning is simplified via non-invasive adjustment of external acoustic conditions, addressing a key limitation of traditional implantable stimulators that require complex internal modulation.

To optimize field delivery in vivo, we systematically investigated two core ultrasound parameters: probe-device distance and excitation power density. Ultrasound attenuation in biological tissues—driven by acoustic impedance mismatches at tissue interfaces (e.g., skull-brain barrier) and distance-dependent energy dissipation—poses a challenge for consistent field generation. As shown in Fig. 3d–e, shortening the probe-device distance from 15 to 5 cm increased the time-domain waveform amplitude by1.8-fold (from 7.2 to 13.2 V/cm), directly correlating with reduced acoustic pathlength and minimized energy loss. This finding underscores the importance of intraoperative probe positioning, where proximity to the device (< 10 cm) allows the MF-SMD device to ensure sufficient energy transmission—even with energy attenuation during conduction through the skull—to reach the 1–3 V/cm (electric field intensity) threshold. Concurrently, within clinically safe power densities (0.2–0.4 W/cm², well below the 0.5 W/cm² threshold for thermal tissue damage set by the FDA), modest power increments yielded proportional gains in output amplitude (Fig. 3f). For example, increasing power density from 0.3 to 0.4 W/cm² elevated the electric field from 8.8 to 11.6 V/cm—this enables the MF-SMD device to still achieve precise tuning within the therapeutic window even after energy transmission is attenuated through the skull.

Based on the above characterization, the principal conclusions of this section are as follows. First, the electrospun PLGA scaffold mimics native dura biomechanics with tunable mechanical properties (ultimate tensile strength: 1.26–4.70 MPa; fracture strain: 16.87–25.44%). Second, the microneedle array structure enhances sonoelectric conversion efficiency by approximately fourfold compared to the planar structure (Fig. S1). Third, the output electric field intensity can be tuned by adjusting the ultrasound excitation power density (linear relationship) and the probe-device distance. For all subsequent in vitro and in vivo experiments, the following optimal parameter set was selected: ultrasound frequency of 40 kHz (providing < 20% skull attenuation while enabling frequency up-conversion to ~ 150–200 kHz output), ultrasound power density of 0.2–0.4 W/cm² (within the FDA safety limit of 0.5 W/cm²), probe-device distance of 5–10 cm (ensuring sufficient energy transmission), and output electric field of 1–3 V/cm at 150–200 kHz (within the established therapeutic window for glioblastoma TTFields).

Safety validation was further reinforced by the device’s inherent design features. The electrospun PLGA scaffold’s biomimetic mechanics (25.44% fracture strain) and porous microstructure prevent mechanical irritation, while the ultrasound-driven mechanism avoids direct electrical contact with neural tissue—eliminating risks of electrode corrosion or current leakage. The electrode (current collector) in our MF-SMD device is made of a gold (Au) thin film (80–100 nm thickness) deposited on a flexible PDMS substrate. Gold is widely recognized as a biocompatible material with excellent corrosion resistance, low toxicity, and good tissue compatibility, and has been extensively used in various implantable device. In conventional TTF systems, extracranial electrodes are in direct contact with the scalp and skin, where prolonged exposure to sweat and the applied electric field can lead to electrochemical corrosion. In contrast, our gold electrode is fully encapsulated within the insulating PDMS layer and PLGA scaffold, preventing any direct contact with body fluids or tissue. This complete encapsulation, combined with gold’s inherent corrosion resistance, effectively eliminates the risk of electrode corrosion. Collectively, these data demonstrate that the MF-SMD achieves triple controllability: (1) clinical-grade electric-field intensity (1–3 V/cm), (2) non-invasive tuning via acoustic parameters, and (3) adherence to safety limits for power and distance. When combined with the scaffold’s morphological and mechanical compatibility with native dura, this establishes a robust physical-dosimetry framework for translating the device to tumor therapy applications.

Overcoming skull-induced electric field attenuation with the MF-SMD

To address the critical challenge of transcranial electric field attenuation in conventional therapies, the MF-SMD device was engineered with an intracranial sonoelectric conversion unit activated directly by external ultrasound, aiming to bypass skull shielding. It is important to distinguish between two types of attenuation in this context. Acoustic energy attenuation refers to the loss of ultrasound energy as it propagates through the skull before reaching the MF-SMD device. Electric field attenuation refers to the loss of the generated electric field as it propagates through tissue. For conventional TTFields, electric field attenuation through the skull is the key limitation. For our subcranially implanted device, the generated electric field does not need to cross the skull again, so electric field attenuation is minimal. The primary parameter of interest in this study is electric field attenuation, as this is the limitation that our device overcomes. To validate that the MF-SMD bypasses skull-induced electric field attenuation, experimental measurements and three-dimensional acoustic-electric simulations were performed. For experimental measurements of skull-induced electric field attenuation, porcine skull specimens were used. The transducer-skull distance was varied from 5 to 15 cm to characterize distance-dependent attenuation properties under controlled conditions. These specimens were isolated and carefully cleaned of overlying skin and soft tissue. Since no skin was present in the ex vivo setup, the measured attenuation represents the intrinsic effect of the skull bone alone. For the schematic illustration shown in Fig. 4a, a commercial medical skull model (ENOVO Brand, China) was used for visualization purposes only.

Fig. 4.

Overcoming Skull-Induced Electric Field Attenuation with the MF-SMD. (a) Schematic of the experimental setup for evaluating skull-mediated electric field attenuation. The schematic uses a commercial human skull model for visualization purposes only; actual ex vivo experiments were performed using porcine skull specimens. (b) Curves showing the effect of varying the source–skull–device propagation distance on voltage amplitude at a frequency of 200 kHz. Data are presented as mean ± SD, n = 3. (c) Amplitude transmission coefficient of the MF-SMD device at different ultrasonic frequencies. Data are presented as mean ± SD, n = 3, Student’s t‑test. *P < 0.05, **P < 0.01, ***P < 0.001. The amplitude transmission coefficient (T) shown in Fig. 4c is defined as T = V_with_skull / V_without_skull. The attenuation percentage is calculated as (1-T) × 100%. The results reveal frequency-dependent attenuation: attenuation was ~ 15% at 28 kHz, increasing to ~ 18% at 40 kHz and ~ 52% at 200 kHz. Note that acoustic energy attenuation through the skull is indirectly reflected in the device’s voltage output, as lower acoustic energy reaching the device results in lower output voltage. (d) Three-dimensional acoustic-electric simulation diagram based on COMSOL. This figure visualizes the acoustic-pressure amplitude distribution at the location of the MF-SMD device. The upper panel is a theoretical comparison (skull absent) and the lower panel simulates conventional transcranial TTFields. (e) COMSOL simulation result diagram of the displacement of the MF-SMD device’s microneedle membrane under ultrasound excitation (0.2 W/cm², 40 kHz). The color scale indicates displacement magnitude: maximum displacement (~ 150 µm, shown in red) occurs at the microneedle tips due to stress concentration, decreasing toward the base (shown in blue). No displacement occurs without ultrasound excitation. With the gap layer set to around 100 µm, the simulated displacement of ~ 150 µm confirms that the device can achieve effective contact-separation motion for tumor electric field therapy.

Under constant drive conditions (transducer center frequency set to 28, 40, or 200 kHz), direct assessment of skull-mediated attenuation was conducted using a porcine skull specimen, positioned between the external transducer and the MF-SMD device secured in a subcranial location (Fig. 4a). When varying the source-skull-device propagation distance at 200 kHz, voltage amplitudes in both skull-present and skull-absent conditions decreased monotonically with increasing distance; however, a consistent separation between the two curves confirmed additional attenuation introduced by the skull (Fig. 4b). The amplitude transmission coefficient, defined as the ratio of voltages with/without the skull, revealed frequency-dependent attenuation: at 28 kHz, attenuation was ~ 15%, increasing to ~ 18% at 40 kHz and ~ 52% at 200 kHz (Fig. 4c). Notably, the MF-SMD device is capable of generating therapeutic-grade alternating electric fields (1–3 V/cm) across all tested frequencies (28–200 kHz), as validated in subsequent in vitro and in vivo experiments. It should be noted that the driving ultrasound frequency (28–200 kHz) is distinct from the generated electric field frequency (100–300 kHz). The device converts low-frequency ultrasound into higher-frequency electric fields via its nonlinear triboelectric mechanism, allowing the use of lower ultrasound frequencies (28–40 kHz) to minimize skull attenuation while still achieving therapeutic field frequencies.

It is important to note that focused ultrasound alone cannot generate tumor treating fields (TTFields). Focused ultrasound produces thermal or mechanical effects (e.g., ablation, sonoporation), whereas TTFields are alternating electric fields that specifically disrupt tumor cell mitosis. Our device is necessary because it converts ultrasound energy into therapeutic electric fields—a function that ultrasound alone cannot perform. Complementary three-dimensional COMSOL Multiphysics simulations further validated these findings by visualizing acoustic-pressure amplitude distributions at the device location, demonstrating reduced pressure attenuation in the presence of the skull at lower frequencies (Fig. 4d). These results establish the MF-SMD’s unique capacity to bypass skull-induced electric field attenuation through two synergistic mechanisms: intracranial transducer placement eliminates extracranial energy dissipation, while low-frequency ultrasound excitation leverages bone’s reduced absorption at 28–40 kHz. Unlike conventional TTF systems, which suffer > 50% transcranial field loss due to extracranial electrodes^53,54, the MF-SMD achieves < 20% attenuation at therapeutic frequencies, directly addressing the fundamental limitation of electric field-based glioma therapies. Further simulations revealed that under an excitation condition of 0.2 W/cm², our device can generate a displacement of approximately 150 µm in the microneedle membrane (Fig. 4e). The displacement is driven by the stress concentration effect at the microneedle tips under ultrasonic excitation. The tapered geometry focuses mechanical stress, resulting in local deformation amplitudes significantly larger than those of a planar film. Therefore, with the gap layer set to around 100 µm, our device is fully capable of performing tumor electric field therapy.

In vitro and in vivo treatment of tumors

The therapeutic potential of the MF-SMD device was evaluated through a combination of in vitro and in vivo studies. In vitro, glioma cells were exposed to intermediate-frequency electric fields generated by activated MF-SMD, and cell proliferation was quantified using fluorescent markers (EdU-488, Alexa Fluor 488, and Hoechst dye) to compare treated cells with untreated controls. Quantitative analysis revealed a marked reduction in tumor cell proliferation in the MF-SMD-treated group, with a proliferation rate of 20% relative to 88% in the control group (Fig. 5a,b), One-way ANOVA revealed a statistically significant inhibitory effect of our device, p < 0.001. Notably, the magnitude of proliferation inhibition observed here is comparable to or exceeds that reported for traditional electric field-based therapies (e.g., tumor-treating fields, TTFields) in preclinical glioma models (with traditional electric field-based therapies achieving approximately 20–50% reduction in tumor burden in glioblastoma cell lines⁵¹ and our study achieving a 68% reduction in tumor burden), highlighting the potential of MF-SMD as a promising alternative.

Fig. 5.

Therapeutic efficacy of MF-SMD in vitro and in vivo. (a) EdU assay analysis of cell proliferation after 12-h MF-SMD treatment; Hoechst-stained cells appear blue, while EdU-stained proliferating cells appear green (Scale = 200 µm). (b) Comparison of proliferation rates between treated and control groups, (n = 5; Data are presented as mean ± standard deviation (SD). One-way ANOVA, p < 0.001). (c) Schematic of in vivo tumor therapy using MF-SMD. (d) Bioluminescence imaging showing tumor suppression by MF-SMD over 28 days of treatment. Images are from the same representative mouse per group at indicated time points. (e) Individual tumor growth curves. Individual tumor growth curves for all mice in the control and MF-SMD groups. Each line represents an individual mouse (n = 4 per group). Tumor burden was quantified by bioluminescence total flux (p/s). The bioluminescence intensity difference between the two groups exceeded two orders of magnitude (> 100-fold) by the end of the treatment period. (f) Kaplan–Meier survival curves for both treated and control groups (n = 4). Statistical analysis was performed using the log-rank (Mantel-Cox) test (P < 0.05). The median survival time for the control group was 37.50 days, while the median survival for the MF-SMD treatment group was not reached by the experimental endpoint.

To validate the in vitro findings, the in vivo therapeutic efficacy of MF-SMD was evaluated using an orthotopic glioma model in C57BL/6 mice (Fig. 5c). Mice with comparable initial tumor burden (assessed by baseline bioluminescence intensity) were randomly assigned to control (device inactive) or treatment (device active) groups. In vivo bioluminescence imaging revealed a divergent trend in tumor progression: the treatment group exhibited a progressive decrease in bioluminescence intensity, indicating reduced tumor burden, whereas the control group showed continuous bioluminescence enhancement, reflecting unchecked tumor growth (Fig. 5d,e). Critically, survival analysis demonstrated a striking therapeutic benefit of MF-SMD: 75% of treated mice survived until the experimental endpoint, compared to 0% survival in the control group (Fig. 5f). Survival analysis was performed using the Kaplan–Meier method with the log-rank (Mantel-Cox) test (P < 0.05). The median survival time for the control group was 37.50 days, while the median survival for the MF-SMD treatment group was not reached by the experimental endpoint (day 50 post-implantation). The single mortality in the treatment group (1 out of 4 mice) occurred on ADay 41 (day 27 of treatment). Postmortem examination revealed no abnormalities directly attributable to the device or ultrasound treatment, and the death was attributed to individual biological variation (e.g., intolerance to anesthesia or surgical recovery), which is occasionally observed in preclinical studies. The remaining three treated mice survived to the endpoint with no signs of distress. These in vivo results confirm the anti-tumor potential of MF-SMD observed in vitro, establishing its ability to suppress glioma growth and extend survival in a preclinical model. The reduction in bioluminescence intensity aligns with the in vitro proliferation data.

Safety evaluation of the sonoelectric artificial meningeal system

To confirm the long-term biosafety of the MF-SMD device for intracranial applications, comprehensive safety evaluations were conducted, encompassing both material-level cytocompatibility and in vivo tissue responses post-implantation. For cytocompatibility assessment, cytotoxicity of device extracts was quantified using the CCK-8 assay, with absorbance at 450 nm measured at 24, 48, and 72 h to evaluate cell viability. No statistically significant differences in absorbance values were observed between the experimental group (exposed to device extracts) and the negative control group across all time points (Fig. 6a,b), indicating negligible cytotoxicity and validating the material’s excellent cytocompatibility.

Fig. 6.

Biosafety assessment of the MF-SMD. (a, b) Cell viability of 3T3 cells cultured with MF-SMD extract or pure medium (control) for 0, 24, 48, and 72 h, as determined by CCK-8 assay. n = 4, Data are presented as mean ± standard deviation (SD). Two‑way ANOVA, P > 0.05. (c) Immunofluorescence staining of cortical tissues at the implantation site (GFAP: green, DAPI: blue, NeuN: red), showing comparable astrocytic activation levels and neuronal density between the implantation group and the control group, indicating that the device did not induce abnormal neural activity (Scale = 100 µm). Tissue samples were collected on day 30 post-implantation after 28 days of daily ultrasound stimulation (treatment group) or sham treatment (control group). (d) Hematoxylin and eosin (HE) staining of scalp tissue and brain tissue at the implantation site (i: scalp tissue, scale = 500 µm; ii: brain tissue, scale = 100 µm), showing no necrosis, hemorrhage, or inflammatory cell infiltration, with histological morphology indistinguishable from controls. For this histological analysis, 3 mice were used per group (n = 3 mice per group). One brain section and one scalp section per mouse were examined. Tissue samples were collected on day 30 post-implantation after 28 days of daily ultrasound stimulation (treatment group) or sham treatment (control group).

In vivo safety was further evaluated in a mouse model, focusing on neural tissue integrity and inflammatory responses. Triple immunofluorescence staining for GFAP (astrocytic activation marker), NeuN (neuronal marker), and DAPI (nuclear counterstain) revealed comparable GFAP expression levels between the implanted and control groups, with no evidence of reactive astrogliosis—an indicator of neural tissue injury (Fig. 6c). Quantitative analysis of GFAP-positive and NeuN-positive cell densities further supported these observations, showing no statistically significant differences between groups (Fig. S2a–b). Additionally, neuronal density and morphological features (e.g., soma size, dendritic arborization) in the cortical region adjacent to the implant were indistinguishable from controls, and qualitative microscopic examination revealed that neuronal cell bodies and dendritic processes appeared morphologically normal with no signs of degeneration (e.g., pyknotic nuclei, vacuolization) , confirming that MF-SMD implantation does not induce detectable neuronal loss, reactive astrogliosis, or tissue inflammation. Histopathological analysis via hematoxylin and eosin (HE) staining of brain parenchyma and scalp tissues at the implantation site further supported these findings: no signs of necrosis, hemorrhage, or inflammatory cell infiltration (e.g., neutrophils, macrophages) were observed in the MF-SMD group, with tissue architecture mirroring that of unimplanted controls (Fig. 6d).

Discussion

This study presents a novel ultrasound-driven, implantable sonoelectric device (MF-SMD) that addresses key limitations of conventional tumor-treating fields (TTF) for glioblastoma. We acknowledge that the required craniotomy compromises the non-invasive nature of pure ultrasound-based approaches. However, this surgical step serves a distinct purpose: to bypass the skull’s high-impedance electrical barrier, enabling consistent delivery of therapeutic electric fields (1–3 V/cm) to the tumor site. This approach differs fundamentally from HIFU ablation or focused ultrasound stimulation, which do not generate TTFields. The initial craniotomy enables long-term, wirelessly powered, localized electric field therapy without continuous external electrodes, eliminating patient compliance burdens. Thus, while not non-invasive, our approach addresses skull-induced electric field attenuation—a limitation that non-invasive ultrasound alone cannot solve. We believe this implantable strategy offers a complementary option for glioblastoma patients, particularly those with superficial or peri-cortical tumors. In a clinical setting, the sonoelectric artificial meningeal device could be intraoperatively placed into the resection cavity after tumor debulking, serving as an adjuvant therapy to ablate residual tumor cells and reduce local recurrence. This approach does not require a separate implantation surgery, as the device leverages the existing surgical window.

The current study used standardized parameters (1–3 V/cm; ~ 150–200 kHz) based on the established TTFields therapeutic window. Patient-specific tumor size and location may require individualized optimization using imaging (MRI/CT) and finite element modeling. In our orthotopic glioma model, tumor cells were implanted ~ 3 mm below the cortical surface. Our device achieved 72% tumor burden reduction and 75% survival, confirming an effective electric field depth of at least 3 mm. However, the maximum penetration depth remains undetermined—a clear limitation.Future studies will characterize depth-dependent field distribution using multi-depth tumor models. For deep-seated tumors, alternative approaches (e.g., stereotactically implanted depth electrodes or multiple devices) may be required.

Key findings demonstrate that MF-SMD integrates a biomimetic PLGA scaffold with multilayer piezoelectric films, enabling wireless ultrasound energy coupling and generating therapeutic electric fields intracranially while bypassing skull-induced attenuation. The primary rationale for evaluating multiple frequencies was to show that lower-frequency ultrasound (28–40 kHz) experiences substantially reduced skull attenuation (~ 15–18%) compared to higher frequencies (200 kHz, ~ 52%). Thus, clinically, lower-intensity ultrasound may achieve the required intracranial field strength while minimizing extracranial energy deposition and side effects. This aligns with the frequency-dependent absorption properties of bone. This attenuation profile, validated via porcine skull specimens and COMSOL simulations, provides a rational basis for selecting lower-frequency excitation to maximize intracranial field strength while minimizing thermal risk.

We acknowledge that our in vivo ultrasound delivery was intermittent due to the direct-contact design with freely moving animals. Nevertheless, observed efficacy confirms that intermittent exposure suffices to activate the device. Future studies will introduce coupling gel and fixed positioning for continuous, uniform energy delivery.

Mechanistically, the electrospun PLGA scaffold mimics native dura biomechanics, ensuring conformal brain adherence and long-term biocompatibility. MF-SMD exhibits no cytotoxicity, reactive astrogliosis, or tissue inflammation. Functionally, in vitro and in vivo models confirmed reduced proliferation (~ 20% vs. ~ 88% in controls) and 75% survival (vs. 0% in controls), establishing proof-of-concept. By merging ultrasound-mediated energy delivery with biomimetic design, MF-SMD overcomes TTF’s technical constraints. Scientifically, it pioneers sonoelectric integration with intracranial implants. Clinically, its dose tunability, wireless operation, and modular placement support personalized strategies for glioblastoma.

These findings validate MF-SMD as a safe, compatible, and potent platform bridging preclinical innovation and clinical need. Future studies optimizing microgap architecture and exploring combinatorial therapies will strengthen translational potential. The PLGA scaffold is designed to degrade over time, but long-term degradation kinetics and chronic device stability remain to be investigated.

We acknowledge several limitations. First, systematic quantification of acoustic and electric field attenuation under controlled conditions was not performed. Second, hydrophone measurements of acoustic field distribution were not performed, as efficacy depends on acoustic energy density reaching the device rather than free-field pressure alone. Third, temperature and cavitation effects were not evaluated; therefore, we cannot claim regarding tissue heating or cavitation. Fourth, the simulation results in Fig. 4d lack experimental validation. Fifth, histological and protein analyses (e.g., Ki-67, cleaved caspase-3) were not examined, though previous studies have documented these effects^55–57. Sixth, skull properties vary across anatomical regions; regional variation was beyond this study’s scope. Seventh, comparative electric field measurements across brain regions were not performed, as the device was placed directly over the tumor. Finally, direct co-culture experiments (tumor cells with healthy neurons) were not performed to demonstrate selectivity. However, our safety data (Fig. 6c, Fig. S2) show no damage to healthy neural tissue, supporting the proliferation-dependent selectivity mechanism. Future studies will address these limitations.

Conclusion

In summary, the MF-SMD exhibited excellent biocompatibility with no adverse post-implantation reactions, and its structural design effectively bypassed skull-induced electric field attenuation, maintaining attenuation below 20% at therapeutic frequencies. It showed robust anti-tumor efficacy in vitro and in vivo, significantly inhibiting GBM cell proliferation and reducing tumor burden. Overall, MF-SMD provides a promising implantable strategy for glioblastoma therapy, strongly supporting clinical translation of localized intracranial electric field treatment.

Methods

Fabrication of electrospun PLGA artificial dura

Poly (lactic-co-glycolic acid) (PLGA; lactide: glycolide molar ratio, LA:GA = 75:25), obtained by bulk polymerization, was used as the base polymer. A mixed solvent of acetone and N,N-dimethylformamide (DMF) was prepared at a 4:6 (v/v) ratio. PLGA powder was added to the solvent to obtain an 18 wt% polymer solution, which was magnetically stirred at 40℃ for 2 h until fully dissolved, yielding a homogeneous 18 wt% PLGA electrospinning solution.

The solution was loaded into a syringe fitted with a metal spinneret and dispensed at 30 µL min⁻¹ using a syringe pump. The spinneret-to-collector distance was set to 12 cm, and a 20 kV potential difference was applied between the spinneret and collector; the ambient temperature was maintained at 25℃. A rotating drum collector was employed, and the winding speed was alternated between 280 and 2800 rpm according to a preset program to produce alternately oriented fiber layers and construct a layered, oriented artificial meningeal membrane. Electrospinning was continued until the membrane reached a thickness of ~ 300 µm, after which collection was stopped. The resultant membrane was left at room temperature to allow residual solvent to evaporate and was stored for subsequent use.

Structure and assembly of the MF-SMD

The device comprised a stack of multilayer functional films, including a ~ 400 µm PDMS microneedle film (primary triboelectric layer) and a ~ 100 µm flexible electrode foil that provided electrical interconnects and solder joints on the internal flexible electrode. The assembly was peripherally insulated by encapsulation, and the bottommost layer integrated an electrospun PLGA meningeal membrane (prepared as described above). After integration, all components formed a single flexible laminate with a total thickness of < 1 mm.

Characterization of electrospun PLGA artificial dura and MF-SMD

The morphology of the electrospun PLGA artificial dura was characterized by scanning electron microscopy (SEM; Phenom Pharos G2). Uniaxial tensile testing of the electrospun PLGA membrane was performed on an Instron 68SC-2 universal testing machine equipped with pneumatic grips. The electrical output of the MF-SMD device was measured and recorded using a Rigol DS2202 oscilloscope with a Rigol RP3300 voltage probe, with the input impedance set to 1 MΩ. Ultrasonic excitation was provided by a commercial transducer driven by a 200 kHz ultrasound generator (Fukoda Ultrasound, China) operating at 20 W electrical input power.

Ex vivo skull attenuation measurement

Ultrasound excitation: A planar ultrasound transducer (Fukoda Ultrasound, China) with a center frequency of 40 kHz (adjustable to 28 kHz and 200 kHz) was driven by an ultrasound generator at a power density of 0.2–0.4 W/cm². The transducer was positioned at distances of 5–15 cm from the skull specimen.

Skull specimen: Porcine skull specimens (thickness 3–5 mm) were used as a human surrogate. The specimens were harvested from the parietal region (cranial vault), a flat and relatively uniform area. This thickness closely approximates human skull thickness (human temporal bone: ~ 1.7 mm; parietal bone: ~ 6–8 mm). The transducer was placed on the outer surface of the skull, and the MF-SMD device was placed on the inner surface of the skull (subcranial location).

Voltage measurement: The output voltage generated by the MF-SMD device was measured using a digital oscilloscope with a 10 × voltage probe (input impedance 10 MΩ). The probe was connected directly to the output electrodes of the MF-SMD device via soldered leads. Voltage waveforms were recorded and analyzed to determine the peak-to-peak amplitude.

Amplitude transmission coefficient: The coefficient T = V_with_skull/ V_without_skull was calculated for each frequency (28, 40, 200 kHz) to quantify skull-induced attenuation.

COMSOL simulation

A 3D acoustic-electric coupled finite element model was constructed using COMSOL Multiphysics. The model included the ultrasound transducer, skull (porcine geometry, thickness 3–5 mm), brain tissue, and the MF-SMD device. The electric field distribution was simulated to evaluate the device performance.

In vitro tumor inhibition study

Primary human glioblastoma cells were isolated from fresh surgical specimens acquired from the Department of Neurosurgery, The First Affiliated Hospital of Sun Yat-sen University. All procedures involving human participants were approved by the Institutional Ethics Committee of the hospital, and written informed consent was obtained from each patient before sample collection. All experiments were conducted in accordance with the approved ethical guidelines. The established primary cell model was used to assess the in vitro antitumor activity of MF-SMD. The device delivered tumor-treating fields (TTFs) at 200 kHz and ~ 2 V/cm. After 24 h of TTF exposure, cell proliferation in the MF-SMD–treated and control groups was assessed. Proliferation was quantified by 5-ethynyl-2′-deoxyuridine (EdU) incorporation, with total nuclei counterstained by Hoechst. Fluorescence microscopy was used to quantify the EdU-positive fraction and determine growth inhibition. For each experimental condition, 5 independent replicate coverslips were prepared. From each coverslip, 3 randomly selected fields of view were imaged, yielding 15 images per group for quantitative analysis. The proliferation rate (percentage of EdU-positive cells) was calculated for each image, and data are presented as mean ± SD. Statistical comparison between groups was performed using one-way ANOVA.

In vivo tumor inhibition study

All procedures involving animals were reviewed and approved by the Animal Ethics Committee of Sun Yat-Sen University (SYSU-IACUC-2023-002678). All animal experiments were strictly performed in accordance with the relevant guidelines and regulations, including the ARRIVE guidelines. To assess the in vivo antitumor efficacy of MF-SMD, an intracranial glioblastoma model was established in C57BL/6 male mice (aged 6–8 weeks, weighing 20 ± 5 g) obtained from the Laboratory Animal Center of Sun Yat-sen University (Guangzhou East Campus Production Area). Mice were anesthetized by intraperitoneal injection of 1.25% avertin (0.2 mL/10 g body weight), placed prone, and secured in a stereotaxic frame to ensure accurate tumor implantation. A 10 µL suspension containing 2.5 × 10⁵ luciferase-expressing GL-261-LUC glioblastoma cells was injected into the right cerebral hemisphere. For tumor tracking, bioluminescence imaging was performed after administration of D-luciferin potassium salt. In the presence of ATP, Mg²⁺, and O₂, luciferase catalyzes the oxidative decarboxylation of luciferin, emitting yellow-green light at ~ 560 nm. An in vivo imaging system (IVIS Lumina Series III, PerkinElmer, USA) was used to monitor bioluminescence intensity as a quantitative measure of tumor burden.

When bioluminescence signals were comparable across animals on ADay 14 (14 days post-tumor implantation), mice were randomized into an untreated control group and an MF-SMD treatment group. The MF-SMD device was positioned subcranially over the tumor to deliver targeted therapy. The treatment regimen consisted of daily continuous ultrasound exposure for 12 h per day. The 12-h daily duration was chosen based on two considerations: (1) clinical relevance—mice have a higher mass-specific metabolic rate and shorter cell cycle than humans, so 12 h/day in mice proportionally corresponds to ≥ 18 h/day in humans as recommended by the EF-14 clinical trial; (2) animal welfare—mice are nocturnal, and the 12-h exposure was scheduled during their active period (dark cycle) to minimize stress. The ultrasound transducer used in this study is a planar transducer (Fukoda Ultrasound, China, 40 kHz). During treatment sessions, the transducer was mounted on a fixed bracket above the cage and pushed/compressed into the cage so that the probe entered the animal’s activity space. Mice moved freely and made direct contact with the transducer surface, enabling wireless activation of the implanted MF-SMD device without any physical wiring. Due to the free-moving nature of the mice, any spatial variation in acoustic intensity across the cage is time-averaged over the 12-h treatment session, ensuring that all animals receive comparable cumulative ultrasound exposure. The working distance between the transducer and the device was approximately 1 cm. The ultrasound power density was set to 0.2 W/cm² (within the clinically safe range below the FDA threshold of 0.5 W/cm² for thermal tissue damage) and stimulation duration of 12 h per day (based on the EF-14 clinical trial). A real photograph of the cage system is provided in Fig. S3. The customized cage has internal dimensions of 13 cm (L) × 9.5 cm (W) × 8.5 cm (H) and is constructed from transparent acrylic. No anesthesia or physical restraint was applied during treatment, allowing ad libitum access to food and water. After each daily treatment, mice were returned to standard housing. The treatment continued until the experimental endpoint (28 days). The 28-day duration covered the full observation period, during which control mice died from progressive tumor burden while treated mice achieved 75% survival, confirming this timeframe is adequate for assessing efficacy. Bioluminescence intensity and survival were recorded longitudinally over a period from ADay 14 to ADay 42(28 days of treatment). The experimental endpoint was defined based on humane endpoint criteria approved by the Animal Ethics Committee of Sun Yat-sen University (SYSU-IACUC-2023-002678), which included: (1) body weight loss exceeding 20% of initial body weight; (2) severe neurological deficits (e.g., inability to ambulate, seizures, or loss of righting reflex); or (3) inability to eat or drink. Animals meeting any of these criteria were humanely euthanized, and the day of euthanasia was recorded as the survival endpoint. The survival experiment was terminated at day 50 because all mice in the control group had died by day 43, and no additional survival events occurred beyond this time point.

Biocompatibility assessment

To evaluate the implant cytotoxicity and tissue safety of MF-SMD, the following experiments were conducted.

Cytotoxicity assay: To evaluate MF-SMD cytotoxicity, 3T3 fibroblasts were used. This choice aligns with the recommendations of the ISO 10,993–5 standard for biocompatibility assessment of medical devices, which advocates the use of established, sensitive cell lines to obtain a standardized and reproducible baseline evaluation of material toxicity. The device extract was prepared as follows: the complete MF-SMD device (total surface area approximately 1 cm²) was immersed in 2 mL of complete DMEM/F-12 culture medium in a sterile centrifuge tube and incubated at 37 °C for 24 h in a 5% CO₂ cell culture incubator. After incubation, the device was removed, and the supernatant was collected and filtered through a 0.22 µm sterile filter to obtain the device extract. The experimental group was exposed to the extract of the MF-SMD device, and the control group received complete culture medium. Cell viability was quantified by measuring absorbance at 450 nm with a multifunctional microplate reader (Synergy H1, BioTek, USA). Measurements were taken at 0, 24, 48, and 72 h post-exposure to assess the time-dependent cytotoxic effects of the extract on 3T3 cells.

Histological staining: On day 30 after device implantation, under deep anesthesia, the brain and overlying scalp were harvested. The treatment group mice received daily ultrasound stimulation (40 kHz, 0.2 W/cm², 12 h per day) for 28 days prior to tissue collection, while the control group mice received sham treatment for the same duration. Brain tissue was cryosectioned (10–20 µm) for immunofluorescence, and adjacent tissue was paraffin-embedded and sectioned (4–6 µm) for hematoxylin and eosin (H&E) staining. Confocal microscopy was used to image regions adjacent to the implant boundary following immunofluorescence, and bright-field microscopy was used to image H&E-stained sections after routine deparaffinization and counterstaining.

Statistical analysis

All data are expressed as mean ± SD. Statistical comparisons were performed using one-way or two-way analysis of variance (ANOVA), with significance defined as *P < 0.05.

Author contributions

B. X. conceived and supervised the project, and determined the research methodology; J. L. performed experiments; J. H., J. R. helped on animal experiments; Y. Q., X. Z., J. H., X. C, and Y. W. helped on device design; X. L. contributed electrical specific guidance;

Funding

No specific funding support was received for this work.

Data availability

Data supporting the findings of this manuscript are available from the corresponding authors upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.