An implantable mechano-electro cascade platform synchronizes neuro-muscle repair Authors: Liping Nan, Wentao Cao, Jiaqi Fang, Pengzhi Shi, Xuehan Jin, Jianguang Wang, Fengshan Jin, Feng Wang, Kaihang Song, Shengfu Liu, Zifei Zhou, Xu Wang, Tianlong Wang, Feng Chen, Junjian Liu, Yun Qian Published: 2026 ABSTRACT AbstractPeripheral nerve injury triggers slow axonal regeneration and irreversible muscle atrophy due to the absence of synchronous neuromuscular repair strategies. We present a fully implantable, self-powered Mechano-driven Electro-adaptable Bioelectronic Implant System (MEBIS) enabling coordinated mechanical-electrical stimulation for integrated neuromuscular regeneration. A programmable robotic actuator delivers quantifiable mechanical massage to denervated muscle, enhancing perfusion while deforming a nanogenerator that converts motion into localized electrical cues. These signals stimulate injured nerves, inducing intracellular Ca2+ elevation and Ca2+-dependent signaling, which activates downstream HIF-1α/AMPK pathways to accelerate Schwann cell activity, angiogenesis, and axonal regrowth. This mechano-electro-biochemical cascade couples muscle preservation with nerve regeneration in a closed loop, overcoming depth limitations and output instability of conventional stimulators. Validated in rat and porcine models, MEBIS significantly improves functional recovery, electrophysiology, and histomorphometry without biocompatibility concerns. Our platform establishes a scalable, feedback-controlled paradigm for precision repair and rehabilitation after peripheral nerve injury.Subject terms: Nanobiotechnology, Biophysics, Trauma, Neurology Introduction Peripheral nerve injuries (PNIs) present major regenerative challenges due to limited neuronal regrowth capacity1. Disrupted electrical signaling after injury impedes neural function and triggers rapid denervation-induced muscle atrophy, compromising motor recovery2,3. Current regenerative strategies, including electrical stimulation (ES), neurotrophic factor delivery, and gene therapy, focus predominantly on axonal regrowth and Schwann cells (SCs) activation4–7. However, they largely neglect a critical consequence of PNI, such as time-sensitive, irreversible atrophy of target muscles8. Without timely reinnervation, denervated muscles undergo progressive mass loss and contractile impairment, culminating in permanent functional deficits9. Once fibrosis or fatty degeneration occurs, motor restoration remains incomplete even with successful nerve regeneration. This disconnect underscores a fundamental limitation of nerve-centric approaches: they fail to address the neuromuscular unit as an integrated system. Thus, therapies enabling simultaneous nerve repair and muscle preservation are urgently needed. Such integrated strategies could alleviate atrophy during prolonged regeneration, establishing a permissive microenvironment for functional recovery2. Recently, some studies reported combinatorial therapies for peripheral nerve repair, particularly ES coupled with physiological activity or physical rehabilitation10–12. For instance, Jin et al. engineered an implantable, respiration-driven physiologically self-regulated ES system that upregulated chemokine receptors (CXCR4/CXCR2), enhanced microvascular density and axonal regrowth, and achieved functional recovery comparable to autografts in rat sciatic nerve defect models12. A parallel study by our work demonstrated that combining treadmill training with ES not only accelerated muscle reinnervation but also increased regenerated myelinated axons versus controls13. These studies underscore the capacity of spatiotemporally synchronized electromechanical stimuli to co-activate regenerative pathways across neural tissues. Despite these advances, critical limitations persist: current platforms cannot deliver precisely integrated mechanical and electrical cues adaptable to regenerative microenvironments. Most lack closed-loop programmability to adjust stimulation parameters in response to tissue feedback, and few simultaneously target muscle preservation alongside nerve repair. To bridge this gap, we introduce a neuromuscular-integrated bioelectronic platform coupling programmable deep-tissue massage with mechano-electric transduction. This self-powered system generates real-time, physiologically tuned mechanical stimulation to mitigate denervation-induced muscle atrophy while converting deformation energy into localized electrical signals for injured nerves, enabling concurrent axonal regrowth and muscle integrity maintenance. Such strategies may overcome key bottlenecks in PNI treatment, advancing functional regeneration toward clinical translation14. Traditional manual massage therapy, a clinically validated biomechanical modality, is a widely used traditional therapeutic method with a long history in clinical practice. It plays an important role in traditional Chinese medicine and is also commonly applied across Southeast Asian countries, including Thailand, Malaysia, and Indonesia. Mechanical massage applies precisely controlled mechanical stimulation to tissues, which not only enhances local blood perfusion and tissue pliability but also promotes metabolic waste clearance. A recent study demonstrated that such mechanical interventions can rejuvenate tissue microenvironments by restoring physiological fluid dynamics and reversing age-associated degeneration14. Beyond its established roles in stress relief and musculoskeletal rehabilitation, recent evidence indicates that massage-mediated mechanotransduction activates regenerative pathways in peripheral nerves and atrophic muscle15–17. These effects arise through modulation of neural excitability and glial responses, and neuroimmune crosstalk regulating inflammation and fibrosis18,19. Critically, its capacity to orchestrate multicellular repair processes positions it as a compelling intervention for integrated neuromuscular recovery. ES augments nerve regeneration by modulating calcium dynamics, neuronal excitability, and SCs plasticity12,20,21. However, conventional ES still presents intrinsic constraints. Implant-based strategies that rely on externally powered electrodes are often associated with tissue fibrosis, charge accumulation, and progressive signal attenuation over time12,22–24. In contrast, non-invasive stimulation approaches generally suffer from limited spatial selectivity and insufficient stimulation strength to effectively target deep or specific tissues25. Critically, neither strategy mitigates denervation-induced muscle atrophy, an irreversible degenerative process that compromises functional recovery despite successful reinnervation26. To address this dual-tissue regenerative challenge, we engineered a fully implantable, self-powered Mechano-driven Electro-adaptable Bioelectronic Implant System (MEBIS). The platform consists of a programmable robotic actuator that provides measurable mechanical massage to denervated muscles, improving perfusion and preventing the progression of atrophy, an MXene-integrated triboelectric-piezoelectric nanogenerator (M-PTNG) that converts mechanical deformation into stable biphasic currents, and a conductive-nanocomposite conduit (PCNC) that guides bioelectric signals to areas of nerve injury (Fig. 1). MEBIS achieves spatiotemporally coupled neuromuscular repair. Mechanical stimulation preserves muscle mass via mechanotransduction, while concurrently generated electrical cues accelerate axonal regrowth through SCs recruitment and neurotrophic secretion. This closed-loop system eliminates stimulation inertia and charge imbalance risks inherent to battery-dependent devices, establishing a translatable paradigm for synchronous neuromuscular functional restoration. Scheme and mechanism of the MEBIS neuromuscular stimulation system. a, b Schematic representation, physical photograph, and structural layout of the MEBIS, illustrating its design, components, and functional configuration. The system consists of three core components: ① A program-controlled massage robot capable of delivering quantifiable and reproducible mechanical stimulation to target muscle tissue; ② A power ①generation module comprising a self-powered MXene-integrated triboelectric-piezoelectric nanogenerator (M-PTNG) that efficiently converts mechanical massage energy into bioavailable electrical signals; and ③ An electroconductive stimulation module comprising PCNC that delivers regulated electrical signals to the injured nerve region, promoting coordinated neural regeneration. c Schematic illustration of the MEBIS enabling functional rehabilitation in paralyzed patients by restoring motor activity. d Schematic illustration of the working mechanism of the MEBIS. I: In the resting state, the massage robot is inactive. No mechanical force is applied, and no electrical output is generated. II: Magnified view of region I, showing no tissue deformation or charge transfer within the power generation module. III: Upon activation, the massage robot applies mechanical force to the muscle surface, inducing localized deformation. Simultaneously, the M-PTNG unit undergoes structural deformation in response to the mechanical input. IV: Enlarged view of region III, illustrating charge displacement and electron transfer within the M-PTNG material. The generated electrical energy is transmitted through conductive leads to the site of peripheral nerve injury, where localized ES is applied to promote nerve regeneration. e Schematic representation of MEBIS-induced cellular signaling. Mechanical cues and electrostimulation delivered by MEBIS lead to an increase in intracellular Ca2+ levels. This elevation may result from a combination of transmembrane influx through putative mechano- or voltage-sensitive channels and/or release from intracellular stores. Elevated Ca2+ subsequently activates downstream signaling cascades, including HIF-1α/AMPK pathways, which modulate gene expression and promote secretion of neuroregenerative and angiogenic factors, thereby facilitating coordinated nerve repair and vascular remodeling. Created in BioRender. Chen, F. (2026) https://BioRender.com/e6pajes. Results Design and integration of the fully implantable battery-free MEBIS To promote peripheral nerve regeneration and functional recovery, we implemented a fully implantable, battery-free MEBIS in vivo. This advanced system comprises three modules: (i) a program-controlled massage robot that quantitatively regulates massage force parameters, (ii) a power generation module based on the M-PTNG device, and (iii) an electroconductive stimulation component: a porous nerve conduit fabricated by 3D printing of poly(ε-caprolactone) (PCL) and functionalized with reduced graphene oxide (rGO) and polypyrrole (PPy) to enable localized ES at the injury site (Fig. 1a, b, Supplementary Movie 1, Fig. 2j–o). Unlike previous studies that were predominantly focused on nerve regeneration, our approach also emphasizes the protection of muscle quality27–29. Moreover, in contrast to conventional ES methods that require external power sources, we employed a multilayer assembly strategy to design a hybrid nanogenerator capable of generating stable electrical power30–32. Design and functional characterization of the power generation module and electroconductive stimulation component. a An electron-cloud-potential-well model is proposed to elucidate the charge transfer between silk and PVDF, both with and without MXene (arrows indicate electron transfer events). b XPS spectrum of the C 1s region of PVDF/MXene film. c Frequency-dependent dielectric constant of PVDF/MXene film with diverse MXene mass fractions. The inset presents the variation trend of the dielectric constant for PVDF/MXene films with diverse MXene mass fractions at 1 kHz. d Optical image and schematic diagram of M-PTNG showing that M-PTNG is composed of multiple layers, such as triboelectric layers (silk and PVDF/MXene), conductive layers (Cu), and encapsulation layers (Kapton and PDMS). e The output voltage of the M-PTNG with diverse MXene mass fractions. f Voltage and current of M-PTNG under different outer load resistance. The output voltage of the M-PTNG under different applied g forces and h frequencies. i Long-term stability tests of the M-PTNG. The working area of M-PTNG was set 16 cm2, and the applied force was 4 N unless otherwise specified. j Light microscopy image of PCNC in lateral view. k, l Low and high magnification SEM images of PCNC in lateral view. m Light microscopy image of the cross-sectional view of PCNC. n, o Low and high magnification SEM images of the cross-sectional view of PCNC. p Porosity of PCL vs PCNC (n = 6 independent samples per group). q Mean pore diameter of PCL vs PCNC (n = 6 independent samples per group). r Contact angle of PCL conduits (n = 3). s Contact angle of PCNC (n = 3). t Conductivity testing results (n = 6 independent samples per group). u Tensile strength testing results (n = 3). Mean values are shown, and error bars represent ±s.d. Statistical analysis was performed using a two-sided Student’s t-test for (p, q) and one-way ANOVA with Tukey’s post hoc test for (t). p values are indicated as *p < 0.05, ***p < 0.001. Experiments in (p–u) were independently repeated at least three times with similar results. The M-PTNG efficiently converts the mechanical energy from massage into electrical energy, producing bidirectional feedback pulses in response to applied stress. After implanting the M-PTNG onto the surface of the gastrocnemius muscle, the PCNC connected via wires is anastomosed to the proximal nerve stump at the injury site (Fig. 1c, d). The synergistic action between the electroactive PCNC and the electrical charges generated by the M-PTNG not only facilitates peripheral nerve regeneration but also, together with the direct mechanical effects of massage, delays muscle atrophy and enhances motor function recovery (Fig. 1). Development of an energy unit and a neural interface To enable efficient electric supply for the ES system, the flexible M-PTNG with high electromechanical conversion capabilities was designed to comprise a silk film as the positive triboelectric layer to lose electrons and a PVDF film as the negative triboelectric layer to receive electrons. Unlike conventional polarization strategies that require complex processing conditions, this study introduced 2D MXene as an anchoring template to induce dipole alignment in PVDF, leading to a significant enhancement in its dielectric properties. To elucidate the mechanism of enhanced output performance of M-PTNG, the electron-cloud-potential-well model was introduced to explain the charge transfer between silk and PVDF, with or without MXene enhancement (Fig. 2a). Here, the electrons residing in specific protein molecular orbits tend to transfer to the empty orbits of PVDF. Moreover, the quantity of transferred electrons is directly related to the disparity in potential-well depths between two materials. Ti3C2Tx MXene, a newly developed 2D nanosheet with abundant -OH surface terminations, enables the long-range growth of TTTT configuration of PVDF and changes the surface electron potential of the triboelectric layer33,34. The increased disparity in potential-well depths between silk and PVDF/MXene leads to a significant enhancement in electron transfer, thereby resulting in a considerable boost in energy output. Considering that the output performance of a TENG was reported to be highly dependent on the transferred charge density, which conversely relied on the surface roughness and dielectric property of the triboelectric material35,36. The PVDF/MXene film with hierarchical spongy porous structure was designed by utilizing a vapor-induced phase-separation approach (Supplementary Fig. 1). The flexible porous PVDF/MXene film would enable higher compressibility and larger deformation at the identical pressure, resulting in greater mechanical energy transformation37,38 (Supplementary Fig. 2). It was worth noting that the Ti3C2Tx MXene nanosheets prepared from a classical chemical etching process could provide numerous hydroxyl (−OH), oxygen (-O), and fluoro (-F) surface terminations (Supplementary Figs. 3 and 4). The abundance of functional groups in the porous PVDF/MXene film facilitated the formation of hydrogen bonds and electron-dipole interactions with the C-H and C-F moieties of the PVDF molecules, leading to robust intermolecular bindings within the PVDF polymer chains. The localized anchoring of PVDF chains on the 2D MXene nanosheets resulted in in situ alignment and orientation of CH2 and CF2 moieties. This transition from initial randomly coiled conformations to long-range all-trans conformation enhanced the macroscopic out-of-plane polarization and piezoelectricity of PVDF/MXene film. As illustrated in Fig. 2b, X-ray photoelectron spectrometer (XPS) spectra demonstrated the presence of oxygen functionalities and hydrogen bonds anchored on the surface of the PVDF/MXene film. The emergence of peaks corresponding to C–O and C=O groups resulting from MXene incorporation indicated the formation of strong interactions, such as hydrogen bonds, between the -OH and -Ox functional groups in the MXene nanosheets and the -CF2 groups within PVDF, in agreement with the previous reports34,39. The strong interactions between PVDF and MXene are associated with an increased ratio of local all-trans conformation in PVDF, with the highest value observed at a MXene mass fraction of 10 wt% (Supplementary Fig. 5a–f). As expected, the dielectric property of PVDF/MXene film showed a similar tendency (Fig. 2c and Supplementary Fig. 5g). This tendency could be attributed to the fact that the appropriate inclusion of conductive MXene nanosheets promoted interfacial polarization between fillers and the polymer matrix, while the excessive introduction of MXene into polymer composites leads to an augmented leaking current and larger dielectric loss. Furthermore, it was worth pointing out that the improved dielectric property endowed the PVDF/MXene film with elevated piezoelectric signals (Supplementary Fig. 5h), which convincingly confirmed the validation and feasibility of our proposed scheme design. After optimizing the dielectric property of PVDF/MXene and integrating this layer with silk nanofiber film (see preparation process in Supplementary Fig. 6), a flexible M-PTNG was devised and characterized (Fig. 2d and Supplementary Fig. 7). The electrical performance of the M-PTNG in response to external stimulation was investigated by utilizing a motor to apply mechanical impulses to the device. Under the synergistic effect of the porous structure and high dielectric property, the 10 wt% MXene incorporated M-PTNG achieved a high output triboelectric voltage of ~160 V and current of ~6 μA (Fig. 2e, Supplementary Fig.8a and Supplementary Table 1). The performance of the device is better than that of the previous implantable neural ES device (Supplementary Table 2). Further investigations of the effective electric power of M-PTNG demonstrated a decrease in current and an increase in voltage with rising load resistances (Fig. 2f), resulting in a peak power density of ~1170 mW m−2 at a load resistance of 80 MΩ (Supplementary Fig. 8b). Furthermore, the M-PTNG exhibited excellent charge injection capacity, as evidenced by its ability to charge a 2.2 µF capacitor to saturation within 80 s (Supplementary Fig. 8c). To explore the reliability, stability, and durability of the M-PTNG, a series of mechanical forces simulating massage pressure (1–4 N) and frequency (2–3.5 Hz) were applied onto the M-PTNG by a modal shaker. The voltage output showed a linear increase with applied force (coefficient of determination R2 = 0.99, slop = 39.6 V N−1, Fig. 2g and Supplementary Fig. 8d) and maintained stability across various frequencies (Fig. 2h and Supplementary Fig. 8e). Meanwhile, the accelerated fatigue test also demonstrated that the output voltage of M-PTNG remained constant even after approximately 10,000 cycles of contact-separation processes (Fig. 2i), indicating good mechanical robustness and reliability for practical scenarios. To evaluate the cytocompatibility of the M-PTNG encapsulated scaffold, SCs were seeded on the surface of the material. Cytoskeletal staining at different time points revealed good cell spreading and elongation compared to the blank control group, indicating favorable cell adhesion and morphology. Live/dead cell viability assays further confirmed the absence of significant cytotoxicity, demonstrating that the material supports SCs survival and proliferation (Supplementary Fig. 9). Using near-field electrohydrodynamic direct-writing (EDW), a porous PCL scaffold was first printed, and subsequently, rGO was dispersed on its surface and in situ complexed with PPy to form the PCNC (Supplementary Fig. 10). Initially, the near-field EDW nozzle extruded a PCL solution under a high electric field, resulting in uniform micron-scale filaments. These filaments were interwoven via the motion of a receiving rod, forming a diamond-grid tubular structure with surface micropores. The scaffold was then immersed in a dispersion containing rGO particles and agitated, enabling rGO adhesion. Following this, pyrrole was introduced to induce the in situ polymerization of PPy on the conduit surface. The process, orchestrated by a reciprocating nozzle, a tungsten steel collector with a defined diameter, and pressurized end devices, gradually deposited the scaffold to the desired thickness under cyclic extrusion and collector rotation (Supplementary Movie 1). The preliminarily-fabricated PCL scaffold exhibited a milky white appearance on both lateral and cross-sectional views, with scanning electron microscopy (SEM) revealing an interwoven, loose fibrous structure (Supplementary Fig. 11a–f). In contrast, optical microscopy and SEM observations of the PCNC displayed a black tubular structure with an interwoven, porous morphology (Fig. 2j–o). Porosity, a critical parameter influencing cell migration, nutrient exchange, and angiogenesis, was estimated at 80–90% for both conduit types, despite slight differences (Fig. 2p)40. Surface micropore diameters ranged from 7 to 9 μm; although the maximum pore diameter decreased somewhat after PPy and rGO incorporation, the pores remained sufficiently large to permit nutrient diffusion while preventing fibroblast infiltration into the lumen that could impede nerve extension (Fig. 2q). Surface hydrophilicity, which is essential for cell adhesion, was markedly improved in the composite41. Pure PCL exhibited an average water contact angle of 89.7°, whereas the PCNC demonstrated an enhanced contact angle of 118.8° (Fig. 2r, s, and Supplementary Fig. 11g). Electrical conductivity measurements indicated that pure PCL had low conductivity; however, following PPy addition, conductivity increased to 0.11 S/cm, and further incorporation of rGO in the PCNC raised the conductivity to 0.13 S/cm (Fig. 2t). Mechanical strain tests confirmed that both conduits possessed high toughness (Fig. 2u), with repeated manual compression showing rapid shape recovery, indicative of excellent resilience for in vivo implantation (Supplementary Movie 2). Both PCL and PCNC exhibited slow and sustained degradation profiles under long-term in vitro conditions. After 5 months of immersion, the cumulative mass loss reached approximately 13.6% for PCL and 6.2% for PCNC (Supplementary Fig. 11h). These results indicate that both materials maintain structural integrity over extended periods, offering stable mechanical guidance channels conducive to long-term peripheral nerve regeneration. The slower degradation observed in PCNC may be attributed to the surface rGO/PPy conductive coating, which likely acts as a barrier to hydrolytic breakdown, thereby enhancing material durability. The hemolysis assay for the implantable nerve conduit nanogenerator demonstrated a hemolysis rate of 1.4%, well below the acceptable threshold, confirming excellent blood compatibility (Supplementary Fig. 11i). Furthermore, live/dead staining of cells cultured on the conduit surface for 3 days showed no significant differences in cell viability among groups (Supplementary Fig. 12a, b). Semi-quantitative analysis using the CCK-8 assay revealed that by day 7, cell viability in the PCNC group was significantly higher than that in both the pure PCL group and the blank control (P < 0.05, Supplementary Fig. 12c). These results suggest that the intrinsic physicochemical properties of rGO and PPy in the PCNC may promote cellular proliferation. Programmable MEBIS for in vivo precise stimulation To enable in vivo neuromuscular stimulation, we developed a customized massage robotic actuator optimized for animal use. This actuator was integrated with a reciprocating platform, pressure sensors, a mechanical force output module, and voltage detection units to construct the MEBIS, as illustrated in Fig. 3a and Supplementary Fig. 13a. The fully assembled MEBIS device was surgically implanted into rats, enabling simulation-based testing of its synchronized mechanical-ES performance. Program-based, precisely controlled massage system: In vivo implantation performance of MEBIS, output characterization, and simulation analysis of electric field distribution and muscle mechanical strain. a Schematic illustration of the MEBIS implantation, intervention, detection, and control system in animal models. b Electrical performance output measurements following mechanical quantification stimulation in animals implanted with MEBIS. c Representative ultrasound images capturing the strain in the tibialis anterior muscle tissue under different compression force stimuli applied by the MEBIS actuator. d Representative strain heatmaps of the gastrocnemius muscle obtained through computational simulations under varying intensities of MEBIS stimulation. e Stress–strain diagrams of the M-PTNG simulated by computer modeling after stimulation at different intensities of MEBIS. f Finite element analysis of the electric field distribution on PCNC under varying MEBIS stimulation intensities. During operation, the reciprocating machine moved rhythmically, causing the latex massage probe at its terminus to contact the rat’s gastrocnemius muscle, and thus inducing deformation. Data from the mechanical sensors and voltage detectors were relayed in real-time to monitoring screens. Concurrently, ultrasonic probes captured imaging data of the muscle deformation throughout the massage process (Fig. 3b, c, Supplementary Fig. 13b, Supplementary Movies 3 and 4). Additionally, throughout the 12-week implantation period, the nanogenerator exhibited stable and robust power generation under a periodic 0.3 N mechanical force, as demonstrated by electrical performance testing (Supplementary Fig. 14). Further analysis was conducted using COMSOL software to dynamically simulate the stress variations experienced by the massage probe, nanogenerator, and muscle tissue, as well as to model the resultant electric field surrounding the PCNC. This simulation provided insights into the MEBIS’s capacity to convert mechanical energy from controlled massage forces into electrical energy via the implanted nanogenerator on the muscle surface, subsequently transmitting this ES to the PCNC to establish a localized electric field (Fig. 3d–f). Programmable ES from MEBIS enhances SCs’ regenerative behavior To assess the biological effects of MEBIS-derived ES on SCs, cells were exposed to the established neural stimulation protocol and compared with unstimulated controls. We assessed the impact of ES on cell morphology, calcium ion expression, neural-specific markers, and neurotrophic factor expression. The results indicated that as the ES intensity increased from 0 to 0.5 V, the expression of neural-specific proteins progressively enhanced. Notably, when the stimulation voltage exceeded 0.5 V, reaching up to 1.5 V, the expression levels of SCs-related proteins progressively declined, indicating a voltage-dependent response and suggesting that 0.5 V represents an optimal stimulation intensity for promoting neural-associated protein secretion (Fig. 4a–d). Based on this observation, subsequent analyses of SCs molecular responses under ES were performed using the identified optimal voltage (0.5 V). We further evaluated the expression of key cytokines involved in peripheral nerve regeneration, including NT3, NGF, BDNF, and VEGF, to elucidate the molecular pathways activated under this stimulation condition. These results demonstrated that ES significantly enhanced the expression levels of these factors (Fig. 4e–h). It is well-known that electrical signals influence the regulation of cellular calcium ion transport, thereby modulating downstream molecular pathways42,43. Calcium ion staining in this experiment revealed that ES notably activated calcium ion channels (Fig. 4i, j). Previous studies have shown that changes in intracellular Ca2+ dynamics are closely linked to the regulation of neural protein expression42. In line with this, our findings indicate that ES influences intracellular Ca2+ signaling, resulting in elevated cytosolic Ca²⁺ levels. Such Ca2+-dependent processes are known to modulate the synthesis and release of neurotrophic factors, thereby supporting neurite extension and neural maturation. Investigating the effects of MEBIS ES on the biological behavior of rat SCs. a, b Fluorescence staining results for S100 and GFAP expression in SCs under different intensities of ES from MEBIS (n = 6). c, d Semi-quantitative analysis of fluorescence staining for S100 and GFAP expression in SCs under varying MEBIS stimulation intensities (n = 6). e–h Quantitative analysis of VEGF, NGF, BDNF, and NT-3 expression in SCs with and without MEBIS-mediated ES (n = 6). i Fluorescence detection of intracellular calcium ion expression in SCs with or without MEBIS stimulation, or at different time durations of MEBIS exposure (n = 6). j Relative intracellular calcium ion expression in SCs under different time durations of MEBIS stimulation, compared to the control group without stimulation (n = 6). Mean values are shown, and error bars represent ±s.d. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test for (c, d) and a two-sided Student’s t-test for (e–h). p values are indicated as *p < 0.05, **p < 0.01, ***p < 0.001. Experiments in (a–j) were independently repeated at least three times with similar results. MEBIS promotes neuromuscular repair following sciatic nerve transection in rats To evaluate the in vivo efficacy of the MEBIS in promoting peripheral nerve regeneration and motor function recovery, we first established a 12 mm sciatic nerve defect model in Sprague–Dawley rats. To evaluate the therapeutic efficacy of our system, animals were randomly assigned to four experimental groups: AUTO (autologous nerve graft), PI (Polymer Implant; non-conductive PCL conduit), NEPI (Nanogenerator and Electroconductive Polymer Implant), and MEBIS groups. In the MEBIS group, a custom-designed, programmable massage probe was used to apply controlled mechanical stimulation to the denervated gastrocnemius muscle. This mechanical input deformed the implanted M-PTNG, thereby generating bioelectric signals through electron transfer. The resulting electrical impulses were delivered through the conductive interface directly to the injured segment of the rat sciatic nerve. In the MEBIS group, massage stimulation was initiated at 2 weeks after surgery and maintained for a total of 10 weeks. Animals received stimulation 5 days per week, with each session comprising two 10-min stimulation periods separated by a 5-min interval. In contrast, the NEPI group received the same nanogenerator and conductive conduit but without massage intervention, enabling us to isolate the effect of mechano-induced stimulation in the MEBIS. Footprint analyses were performed at 4, 8, and 12 weeks for sciatic functional index (SFI) scoring, and electrophysiological evaluations, along with tissue collection, were conducted at 12 weeks (Fig. 5a). Detailed implantation procedures are provided in Supplementary Fig. 15. In vivo evaluation of MEBIS implantation in SD rats for nerve regeneration and motor function recovery. a Schematic illustration of the key concept of MEBIS animal experiments. Created in BioRender. Chen, F. (2026) https://BioRender.com/2m7raag. b The results of HE, TEM, and neuro-specific immunofluorescence staining of the regenerated sciatic nerve in rats 3 months after the operation (n = 6). c–e Evaluation of nerve regeneration at 3 months post-surgery, including myelinated neuron density, number of myelin sheaths, and nerve myelin sheath thickness (n = 6). f–i Semi-quantitative statistical results of immunofluorescence of S100, MBP, Tuj1, and NF200 in the regenerated sciatic nerve of rats 3 months after the operation (n = 6). j–l Evaluation of muscle atrophy and recovery at 3 months post-surgery. j Upper panel: gross morphology of bilateral gastrocnemius muscles (left: injured side; right: uninjured side) from each group. Lower panel: corresponding HE staining images of muscle tissue sections (n = 6). k Quantification of gastrocnemius muscle wet weight (n = 6). l Semi-quantitative analysis of average muscle fiber diameter(n = 6). Mean values are shown, and error bars represent ±s.d. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test for (c–i). p values are indicated as *p < 0.05, **p < 0.01, ***p < 0.001. Experiments in (b–i) were independently repeated at least three times with similar results. Three-dimensional CT imaging performed 3 months post-surgery confirmed that both the PCNC and the M-PTNG maintained their designated positions, indicating the system’s long-term stability in vivo (Supplementary Fig. 16). Histological evaluation using HE and toluidine blue (TB) staining, along with transmission electron microscopy (TEM) analysis, was employed to assess nerve regeneration at 12 weeks post-surgery. The PI group displayed sparse and disorganized nerve fibers with extensive fibrous tissue infiltration, and SEM analysis revealed that the density of myelinated axons, the number of myelin sheaths, and myelin thickness were significantly reduced compared to the AUTO group. The NEPI group exhibited notable improvements relative to the PI group, likely owing to the intrinsic electrical and chemical properties of the conductive nerve conduit and the generation of physiological electrical signals during voluntary activity. Importantly, the nerve morphology, TEM ultrastructure, and quantitative analyses in the MEBIS group were comparable to those in the AUTO group, indicating effective nerve regeneration. Furthermore, immunofluorescence staining for nerve-specific proteins (MBP, S100, NF200, and Tuj1) and subsequent semi-quantitative analyses corroborated that the regenerative outcomes in the MEBIS group were similar to those observed in the AUTO group (Fig. 5b–i, Supplementary Fig. 17a). Functional recovery of hindlimb muscles following sciatic nerve transection We comprehensively evaluated functional recovery over a 12-week period following sciatic nerve transection. Initial toe-spread reflex scores were uniformly low across all groups. The reflex, a sensitive indicator of hindlimb functional restoration, was scored on a scale from 0 to 2 (0 = no spreading, 1 = partial spreading, 2 = full spreading). In the PI group, scores gradually improved between weeks 4 and 8 and plateaued thereafter. In contrast, MEBIS-treated mice exhibited a progressive increase in toe-spread scores throughout the 12-week period, achieving significantly higher values than PI animals and approaching the levels observed in the AUTO group, indicating enhanced neuromuscular recovery (Supplementary Fig. 17e, f). Isometric muscle force assessment Isometric contractile force of the tibialis anterior (TA) muscle was measured in vivo to quantify functional muscle recovery. Animals were anesthetized, and the foot of the injured limb was secured to a motor-driven footplate connected to a force transducer. Subcutaneous electrodes delivered ES to the medial and lateral TA, and the resulting torque was quantified as muscle-generated force (Supplementary Fig. 17g). At 12 weeks post-surgery, the MEBIS group demonstrated significantly greater force generation compared with the PI group, confirming that enhanced reinnervation translated into functional improvements in muscle strength. Sciatic functional index and electrophysiology SFI was calculated at 4, 8, and 12 weeks based on footprint parameters, including toe spread, print length, and intermediate toe spread. Initial SFI values were low in all groups, but by 12 weeks, both MEBIS and AUTO animals showed substantial recovery, significantly outperforming the PI and NEPI groups (Supplementary Fig. 17k). Complementary electrophysiological assessments at 12 weeks revealed that CMAP amplitudes and nerve conduction velocities (NCV) in the MEBIS group were comparable to those in the AUTO group, while PI animals exhibited marked deficits (Supplementary Fig. 17b–d). Gait and locomotor performance Gait analysis using the CatWalk XT system corroborated these findings. At 12 weeks post-injury, MEBIS-treated rats exhibited increased stride length, higher average paw contact intensity, faster swing velocity, and reduced swing duration relative to PI and NEPI groups, approaching values observed in AUTO animals. Open field testing further demonstrated that MEBIS rats traveled greater distances and frequently explored the central area, indicating superior locomotor activity and exploratory behavior (Supplementary Fig. 17h–j). Footprint intensity maps illustrated enhanced gait symmetry and coordination (Supplementary Fig. 18). Muscle morphology and wet weight Gross examination of the gastrocnemius muscle revealed severe atrophy, reduced fiber quality, and inflammatory infiltration in PI animals. In contrast, muscles from the MEBIS and AUTO groups maintained improved morphology. Quantitative analysis of muscle wet weight confirmed superior preservation of muscle mass in these groups. Histological assessment via H&E staining showed that PI muscles contained smaller fibers with extensive fibrosis, whereas the MEBIS group displayed fiber architecture comparable to that of AUTO animals. The NEPI group exhibited partial improvement, yet remained inferior to MEBIS, underscoring the critical role of massage in muscle preservation (Fig. 5j–l). Collectively, these results indicate that MEBIS treatment effectively enhances functional, electrophysiological, and structural recovery of denervated muscles, producing outcomes comparable to those achieved with autograft repair. FG retrograde tracing reveals enhanced neuronal connectivity in MEBIS-treated animals To evaluate the extent of axonal regeneration and reinnervation, we performed FG retrograde tracing in spinal cord motoneurons and DRG sensory neurons at 12 weeks post-injury. Representative immunofluorescence images showed clear FG labeling in the ventral horn motoneurons and DRG sensory neurons across all experimental groups (Supplementary Fig. 19a). Quantitative analysis revealed that the number of FG-positive motoneurons and DRG neurons was significantly higher in the MEBIS group compared with the PI and NEPI groups (Supplementary Fig. 19b, c). The FG-labeled neuron counts in the MEBIS group approached levels observed in the AUTO (autograft) group, indicating robust regeneration and re-establishment of functional neuronal connectivity. These results suggest that MEBIS treatment effectively promotes axonal outgrowth and supports the recovery of both motor and sensory pathways following peripheral nerve transection. MEBIS functional and molecular role in nerve regeneration To elucidate the potential mechanisms by which the MEBIS promotes neural functional recovery, we performed RNA-seq on regenerated nerve tissues obtained from different treatment groups of SD rats at 1 month post-surgery. After MEBIS normalization, both the volcano plot and heatmap analyses identified 398 differentially expressed genes, including 357 upregulated and 41 downregulated genes (Fig. 6a, b). Subsequent GO analysis revealed significant enrichment of these genes in categories such as biological regulation, response to stimulus, and transporter activity (Fig. 6c). Furthermore, KEGG pathway analysis demonstrated that the differentially expressed genes were involved in key signaling pathways, including calcium signaling, AMPK, and HIF-1 pathways (Fig. 6c, d). These pathways appear to act synergistically in mediating the effects of ES on nerve regeneration. Exploration of gene expression patterns and functional enrichment analysis following nerve regeneration in animal models. a, b Volcano plot and heatmap of differentially expressed genes (DEGs), highlighting upregulated (red points) and downregulated (blue points) DEGs. Differential expression analysis was performed using a two-sided statistical framework with multiple testing correction applied using the Benjamini–Hochberg method (n = 3). c Gene Ontology (GO) enrichment analysis, including Biological Processes (BP), Cellular Components (CC), and Molecular Functions (MF) (n = 3). d KEGG pathway enrichment analysis of DEGs from the MEBIS group compared to the blank control group (PI), visualized as a bubble plot. KEGG enrichment analysis was conducted using a two-sided hypergeometric test with FDR correction (n = 3). e Gene Set Enrichment Analysis (GSEA) of nerve regeneration-related signaling pathways and biological processes (n = 3). f GSEA of osteogenesis-related signaling pathways and biological processes. Gene set enrichment analysis (GSEA) was performed using permutation-based testing with FDR adjustment (n = 3). g Western blot analysis of the core protein expression involving nerve regeneration, Calcium, AMPK, and HIF-1 signaling pathway between PI and MEBIS groups (n = 3). Experiments in (c–g) were independently repeated at least three times with similar results. All statistical tests were two-sided unless otherwise specified. Clustering analysis of the differentially expressed genes within these pathways revealed a predominant upregulation in the MEBIS-treated group, suggesting that MEBIS activates these pathways to exert its neuroregenerative effects (Fig. 6e). Gene Set Enrichment Analysis (GSEA) further confirmed the activation of the neural system and the modulation of receptor–ligand interactions and transporter activities associated with neural activation (Fig. 6f). Western blot validation showed significant activation of key molecules in the calcium, AMPK, and HIF-1 signaling pathways following MEBIS treatment. Moreover, both protein and gene expression analyses of neural regeneration markers indicated that MEBIS treatment upregulated these indicators, further substantiating its potent therapeutic effect on neural functional recovery (Fig. 6g and Supplementary Fig. 20). Considering that nutrient supply is critical for tissue regeneration and depends on adequate vascular perfusion, we also assessed neovascularization around the regenerated tissue44. Our previous studies demonstrated that conductive nerve conduits promoted angiogenesis, thereby facilitating nerve tissue regeneration, and that massage inherently enhanced blood circulation45. Immunofluorescence staining for CD34 and CD31, along with semi-quantitative analysis, reveals that MEBIS significantly promotes neovascularization, thereby greatly enhancing nutrient delivery and creating a favorable microenvironment for nerve regeneration (Supplementary Fig. 21). MEBIS enhances neurotrophic factor expression in distal nerve stumps To determine whether the neurotrophic responses observed in isolated SCs cultures are recapitulated within the complex in vivo regenerative environment, we quantified protein levels of key neurotrophic factors in whole distal sciatic nerve stumps at 4, 8, and 12 weeks after sciatic nerve injury. ELISA-based analyses were performed to assess VEGF, NGF, BDNF, and NT-3 expression across all experimental groups. When normalized to the AUTO group, neurotrophic factor levels exhibited distinct temporal patterns across treatment groups (Supplementary Fig. 22). At early and intermediate time points (4 and 8 weeks), MEBIS-treated nerves consistently exhibited higher expression levels of NGF, BDNF, NT-3, and VEGF than those in the PI and NEPI groups, indicating enhanced neurotrophic support during the active phase of regeneration. By 12 weeks post-injury, neurotrophic factor levels in the MEBIS group remained elevated relative to PI and NEPI controls, while showing a trend toward stabilization, consistent with a more advanced stage of regenerative remodeling. These in vivo protein measurements integrate contributions from SCs, regenerating axons, vascular components, and immune-derived signals within the distal nerve stump, providing physiologically relevant evidence that MEBIS sustains a pro-regenerative molecular environment during peripheral nerve repair. In vivo biocompatibility evaluation Biocompatibility is critically important for implantable devices. We evaluated the tissue response by performing immunofluorescence staining for CD68 and TNF-α on tissues surrounding the implanted device at various time points. These results indicated that significant inflammatory cell infiltration was observed during the first to third week post-implantation; however, by 4 weeks, the inflammation had almost completely subsided, and no obvious inflammatory infiltration was observed at 8 weeks (Supplementary Fig. 23a, b). Furthermore, HE staining of major organs (heart, liver, spleen, lung, and kidney) collected from SD rats at 12 weeks post-implantation revealed no significant pathological changes, confirming the excellent biosafety of the system and its suitability for long-term implantation and subsequent applications (Supplementary Fig. 23c). These findings underscore the potential clinical translational value of the device. MEBIS promotes neuromuscular repair in a mini-pig injury model Inspired by the promising outcomes observed in cell-level and SD rat studies, which demonstrated that the MEBIS effectively promoted peripheral nerve regeneration and functional recovery, we further evaluated its therapeutic potential in a Bama mini-pig sciatic nerve defect model, highlighting its potential for clinical translation. In this model, following similar surgical procedures to those employed in SD rats, the sciatic nerve was exposed and transected to create a long-gap defect. The prepared PCNCs were sutured to the nerve stumps, and the M-PTNG device was implanted on the ipsilateral gastrocnemius muscle (Fig. 7a and Supplementary Fig. 24). Evaluation of MEBIS effects in the Bama mini-pig sciatic nerve injury model. a Schematic representation and outcome of gait analysis in Bama mini-pigs with sciatic nerve defect, before and 6 months after surgery, with or without MEBIS intervention. Postoperative pigs in the PI group exhibited persistent hindlimb paralysis and abnormal gait, whereas the MEBIS-treated group showed marked recovery of motor function and near-normal gait restoration. b Representative color Doppler ultrasound images of the gastrocnemius muscle cross-section at 6 months post-surgery. c Immunofluorescence staining of fast and slow muscle fibers in the gastrocnemius muscle. d Immunofluorescence staining of NF200 and Tuj1 in regenerating sciatic nerve fibers. e Semi-quantitative analysis of average muscle fiber diameter (n = 6). f Semi-quantitative analysis of relatively fast fiber expression. g Semi-quantitative analysis of relative slow fiber expression (n = 6). Semi-quantitative analysis of Tuj1 (h) and NF200 (i) expression in regenerated sciatic nerves of Bama mini-pigs at 6 months post-operation (n = 6). Mean values are shown, with error bars representing ±s.d. Statistical analysis for (e–i) was performed using one-way ANOVA followed by Tukey’s post hoc test. p values are indicated as *p < 0.05, **p < 0.01, ***p < 0.001. Experiments in (c–i) were independently repeated at least three times with similar results. Postoperative assessments included ultrasonic evaluations of muscle status, gait analysis via video recordings, and histological as well as TEM analyses of both the gastrocnemius muscle and regenerated nerve tissues. Gait analysis revealed that the MEBIS group exhibited significant improvement in limping compared to the PI group (Fig. 7a and Supplementary Fig. 25). Ultrasound imaging of the gastrocnemius muscle indicated that while the PI group displayed disorganized, thinned, and irregular muscle fibers, the MEBIS group showed recovery in muscle fiber diameter and a more regular morphology (Fig. 7b and Supplementary Fig. 24d). Furthermore, immunofluorescence staining for fast and slow muscle proteins demonstrated that the MEBIS group had a significantly higher proportion of muscle proteins associated with contraction relative to the PI group. Consistently, HE and Masson’s trichrome staining revealed that the PI group had marked muscle fiber atrophy and increased fibrotic tissue, whereas the MEBIS group maintained larger fiber diameters (Fig. 7c, e–g, Supplementary Fig. 26a, c). Similarly, analyses of the sciatic nerves using immunofluorescence for neural-specific proteins, HE, TB staining, and TEM demonstrated that, regardless of the parameters assessed such as neural-specific protein expression, nerve fiber count, morphology, number of myelin sheaths, axon diameter, and myelin thickness, the results in the MEBIS group were comparable to those observed in the SD rat experiments (Fig. 7d, h, i, Supplementary Figs.26b, d, e and 27). Collectively, the results from both small and large animal models confirm that the MEBIS significantly enhances peripheral nerve regeneration and functional recovery, thereby underscoring its promising potential for clinical translation. Discussion PNI disrupts the transmission of physiological electrical signals to target muscles, leading to denervation-induced muscle atrophy and impaired motor function. Effective treatment requires coordinated structural and functional restoration of both nerves and muscles, a process termed neuromuscular integrated repair46. ES has emerged as a critical modality for promoting peripheral nerve regeneration by accelerating axonal growth, activating SCs that support nerve repair, and upregulating neurotrophic factors, thereby significantly enhancing regenerative outcomes. Massage, as a non-invasive physical therapy, mechanically stimulates muscles and surrounding tissues to improve local blood circulation, activate the neuromuscular system, prevent muscle atrophy, and facilitate functional recovery. Its non-invasive, safe, and easily operable nature makes it an ideal adjunct therapy for muscle dysfunction47,48. Although both ES and massage show promise for nerve regeneration and muscle protection independently, no mature technology currently exists to integrate these approaches synchronously18,49–51. Developing truly neuromuscular-integrated repair strategies remains a critical frontier for future therapeutic innovation. Recent mechano-electrical-biochemical cascade systems utilize mechanical force to generate electrical stimuli, which in turn trigger extracellular and intracellular signaling. This process enhances calcium-mediated mechanotransduction. These cascades stimulate cell proliferation, migration, and angiogenesis, and support tissue repair, all while avoiding the risks associated with conventional electrical stimulators, such as secondary tissue damage and limited portability52. Our approach aligns with these emerging concepts while introducing key innovations. Unlike previous self-powered platforms that concentrate solely on neural regeneration, our MEBIS provides coordinated stimulation to both nerve and muscle simultaneously to prevent early denervated muscle atrophy and limb dysfunction. In this study, we developed a fully implantable, self-powered MEBIS that integrates program-based, precisely controlled massage with localized ES for comprehensive neuromuscular repair53. The MEBIS comprises three core components: (1) a programmable massage robot capable of delivering quantifiable, tunable mechanical force to denervated muscles; (2) a power generation module comprising an M-PTNG, which converts mechanical energy into electric signals; and (3) an electroconductive stimulation module that transmits the generated signals to the site of nerve injury. This closed-loop system couples mechanical muscle stimulation with simultaneous, localized ES to the nerve, thereby facilitating both muscle preservation and axonal regrowth54–56. This neuromuscular-integrated bioelectronic platform is coupled with programmable deep-tissue massage for precise mechano-electric stimulation. Unlike ultrasound-triggered piezoelectric systems, which suffer from depth-dependent energy loss and imprecise stimulation, our MEBIS utilizes direct mechanical force conduction through a programmable massage actuator to deliver lossless, quantifiable pressure to target muscles13,57,58. This precise mechanical input not only enhances local perfusion and metabolic activity within atrophic muscles but also deforms an embedded nanogenerator to produce stable electrical signals. By routing these bioelectric cues to adjacent nerve injury sites, MEBIS simultaneously modulates the electrochemical and biochemical microenvironments of both muscle and nerve. Stimulating nerves without synchronous muscle modulation risks a mismatch between nerve regeneration and muscle recovery, ultimately hindering the restoration of motor function59. In effect, the system is designed such that a single mechanical intervention can simultaneously support muscle maintenance through mechanobiological mechanisms and facilitate nerve regeneration via mechano-electrical and biochemical signaling. This integrative strategy overcomes the electrical output and targeting limitations of traditional piezoelectric materials, enabling a transformative shift from single-mode nerve stimulation to bidirectional “muscle on nerve” coordinated stimulation. To our knowledge, no prior system integrates a tightly controlled, parameterized massage platform with a self-powered electrical stimulator to achieve such coordinated, organ-spanning repair. This dual-modality approach overcomes the penetration and reproducibility limitations of traditional methods, offering a highly controllable strategy for deep neuromuscular core regeneration and paving the way for clinically translatable adaptive rehabilitation devices. The MEBIS platform integrates mechanical sensors and real-time voltage detectors to achieve stable, precisely regulated mechano-electrical conversion. Precise, gentle non-invasive mechanical interventions have been shown to reverse age-related tissue decline14. Unlike conventional systems, which often produce fluctuating outputs and lack accurate control, our MEBIS continuously monitors the force-to-voltage process and adjusts stimulation parameters on the fly. To quantify how these electrical signals influence SCs behavior, we applied a range of voltages to SCs cultures and measured expression of S100 and GFAP. The results showed that when the ES intensity increased from 0 to 0.5 V, the expression of neural-specific proteins was progressively enhanced in a positively-nonlinear-correlation manner. Consequently, we confirmed the quantitative relationship of the mechano-electrical-biochemical transduction by MEBIS for neuromuscular repair. Notably, SCs-associated protein expression was diminished as the applied voltage increased beyond 0.5 V, revealing a clear voltage-dependent response. Accordingly, we selected 0.5 V as the optimal stimulation intensity to maximize neural-associated protein secretion. Previous studies demonstrated ES facilitated nerve regeneration by activating various intracellular signaling pathways, including PI3K/Akt and MAPK60,61. Among them, the influx of calcium ions (Ca2+) is widely recognized as a critical secondary messenger in mediating cellular responses to electrical signals62. Ca2+ signaling regulates numerous cellular processes, including neuronal excitability, cytoskeletal remodeling, and neurotrophic factor expression63,64. In the context of MEBIS, mechanical feedback generated by the massage actuators applies rhythmic compressive forces to the denervated muscle surface. This mechanical deformation is transmitted to the implanted nanogenerator, where mechano-induced charge transfer occurs, converting mechanical energy into localized bioelectrical signals. The resulting electron flow produces consistent, spatially confined electrical cues that are delivered to the injured nerve microenvironment. These cues improve intercellular communication and trigger downstream biochemical and molecular responses, thereby initiating the cascade from electrical to biological signaling. Building on these findings, our combined in vitro and in vivo studies demonstrated that MEBIS not only elevated intracellular Ca2+, thereby enhancing signal transduction critical for neurite extension and SCs activation, but also exerted broad transcriptional effects in a physiologically relevant setting. Specifically, RNA-seq of regenerated rat sciatic nerve tissue revealed robust upregulation of HIF-1α and AMPK signaling components: HIF-1α target genes involved in VEGF-mediated angiogenesis, glucose transporters, and erythropoietic factors; and AMPK-regulated genes controlling autophagic flux, mitochondrial biogenesis, and glycolytic enzymes65–67. This dual activation likely creates an energy- and oxygen-optimized microenvironment that accelerates debris clearance, supports glial proliferation, and sustains axonal metabolic demands. Functionally, we observed enhanced capillary density and accelerated myelin sheath restoration in MEBIS-treated nerves, correlating with improved motor recovery. Importantly, MEBIS delivers consistent, adjustable electrical and mechanical cues, allowing us to fine-tune stimulus parameters for individualized treatment regimens. Together, these results underscore MEBIS’s capacity to coordinate multiple regenerative processes: vascularization, metabolic support, and cytoskeletal remodeling, offering a powerful, integrated approach to peripheral nerve repair. Although the present study focused on functional and histological outcomes up to 12 weeks after implantation, the long-term durability of the MEBIS platform remains to be fully established. Prior studies on implantable self-powered bioelectronic systems provide a supportive context: implantable piezoelectric and triboelectric nanogenerators with comparable working principles and constituent materials have demonstrated stable electromechanical performance, preserved stimulation efficacy, and favorable biocompatibility over extended implantation periods, typically ranging from several weeks to months12,68–71. Notably, these studies reported that self-powered neuromodulation devices can maintain consistent electrical output with minimal chronic inflammatory responses at the tissue-device interface. In addition, flexible polymer encapsulation strategies, such as PDMS- and parylene-based coatings, have been used in implantable bioelectronic devices to provide biocompatible encapsulation and structural protection of the device core. In the present study, robust functional recovery was observed at the 12-week endpoint, providing initial evidence for the therapeutic efficacy of MEBIS in vivo. Our evaluation in both small (Sprague Dawley rat) and large (Bama mini pig) animal models demonstrated excellent biocompatibility, mechanical integrity, and positional stability of the device. Histological analysis of major organs from treated rats revealed no significant toxicity or pathological abnormalities. Over the 12-week observation period, no adverse inflammatory responses or foreign-body reactions were detected. Mechanically, the robotic massage probe maintained consistent force output, and the device assemblies remained securely anchored under dynamic weight-bearing conditions. Functionally, treated animals exhibited notable improvements in locomotor scoring and electrophysiological parameters compared with controls. Muscle preservation was evident, with the gastrocnemius muscle retaining mass and contractile integrity, and tissue histology showing denser axonal networks and more developed myelin sheaths. Large-animal studies yielded consistent findings, as mini-pigs receiving MEBIS regained coordinated gait significantly earlier than untreated counterparts. Importantly, unlike prior strategies targeting neural or muscular components in isolation, the MEBIS platform integrates direct mechanical massage to the denervated musculature with in situ-generated ES delivered to the injured nerve. This combined approach mitigates atrophy, preserves microvascular perfusion, and promotes axonal regrowth, yielding synergistic benefits in neuromuscular repair. Collectively, these data underscore the clinical translational potential of MEBIS as a self-powered rehabilitation platform capable of accelerating recovery of motor function. However, we acknowledge that true long-term durability, beyond 12 weeks, requires follow-up studies extending to at least 6 months or more, which would provide definitive evidence for sustained efficacy. To contextualize the advantages of MEBIS relative to existing neuromuscular stimulators and rehabilitation strategies, both mechanistic and practical aspects must be considered. Traditional neuromuscular stimulators typically rely on externally applied electrical pulses delivered through tethered devices or repeated interventions, which can result in variable stimulation efficiency and limited engagement of the denervated target tissue. By contrast, MEBIS is a fully implantable, self-powered system that converts mechanically applied massage directly on the denervated muscle into localized electrical cues. This approach enables continuous, autonomous, and site-specific stimulation, providing precise spatiotemporal control over the stimulation parameters while simultaneously activating multiple regenerative processes, including SCs activation, angiogenesis, and metabolic support. Importantly, by coupling direct muscle stimulation with in situ electrical cues delivered to the injured nerve, MEBIS establishes a closed-loop repair paradigm that integrates muscle recovery with nerve regeneration, addressing two critical aspects of functional restoration concurrently. This design represents a conceptual and practical innovation beyond current strategies. Unlike conventional rehabilitation methods, which largely provide passive or generalized mechanical therapy, MEBIS delivers a reproducible, tunable mechano-electrical intervention that can be translated to larger animal models and, ultimately, patients. Together, these mechanistic and translational advantages underscore the unique therapeutic potential of MEBIS in accelerating coordinated neuromuscular repair and functional recovery following PNI. In summary, we present a fully implantable, self-powered MEBIS that synergistically combines programmable mechanical massage with mechano-electrical transduction to promote neuromuscular regeneration. The integrated system, comprising a programmable massage robot module, power generator module (M-PTNG), and electroconductive module (PCNC), demonstrated efficient energy conversion and stable electrical output in vitro, while enhancing SCs activity and neuroregenerative factor expression. In rodent models of sciatic nerve injury, MEBIS achieved functional recovery comparable to autografts, as evidenced by improved nerve fiber density, myelin thickness, and muscle morphology restoration. Molecular analyses identified calcium signaling, AMPK, and HIF-1α pathways as key mediators of this regenerative process. The system maintained excellent biocompatibility across both rodent and porcine models, with no observed inflammatory responses or organ pathology. Collectively, our findings suggest that the MEBIS represents an innovative and efficient strategy for the synergistic treatment of PNIs. Limitations and future perspectives An important consideration for the potential clinical translation of the MEBIS system is the use of non-biodegradable components and the possible need for secondary surgical intervention. Notably, the MEBIS platform consists of two functionally and anatomically distinct implanted components. The first component is the nerve-interfacing ES conduit, which remains in situ and is not intended for removal. This conduit is primarily composed of PCL with low concentrations of rGO and PPy. In our previous studies, rGO exhibited good biocompatibility, and this observation is further supported by numerous high-quality reports demonstrating that both rGO and PPy are well tolerated and suitable for long-term biomedical applications72–77. Given the relatively low content of non-degradable conductive additives and the risk of secondary injury associated with removing materials integrated with regenerated neural tissue, this component is deliberately designed to remain implanted. In contrast, the second component, the mechano-electric generator, is positioned superficially on the muscle surface within the subcutaneous layer rather than embedded within deep muscle or neural tissue. Given its small size relative to the host tissue, the device can be removed at the end of the treatment period with minimal surgical intervention. Specifically, removal would require only a small superficial incision, disconnection of the lead linking the generator to the stimulation conduit, and gentle separation of surrounding subcutaneous tissue, without disruption of regenerated nerve fibers or deep muscle structures. Therefore, from a structural and anatomical standpoint, this retrieval process is expected to impose minimal mechanical disturbance to regenerated neuromuscular tissues. While we hypothesize that such a localized and superficial removal of the device would allow rapid postoperative recovery, this assumption has not yet been directly validated by dedicated functional or histological assessments following device removal. Future studies will therefore focus on systematically evaluating tissue recovery dynamics after device removal, as well as further optimizing device miniaturization and exploring biodegradable or bioresorbable material alternatives, to enhance the long-term practicality and translational feasibility of the MEBIS platform. Methods Antibodies The following primary antibodies were used in this study (supplier, catalog number, application, dilution): Primary antibodies for immunofluorescence (IF): Anti-S100 (Proteintech, 15146-1-AP; IF, 1:200) Anti-GFAP (Proteintech, 16825-1-AP; IF, 1:200) Anti-MBP (Proteintech, 10458-1-AP; IF, 1:200) Anti-III Tubulin (TUJ1) (Proteintech, 66375-1-Ig; IF, 1:200) Anti-NF200 (Proteintech, 18934-1-AP; IF, 1:200) Anti-CD34 (Servicebio, GB111693-100; IF, 1:200) Anti-CD31 (Servicebio, GB11063-2-100; IF, 1:200) Anti-CD68 (Servicebio, GB113109-100; IF, 1:200) Anti-TNF-α (Servicebio, GB11188-100; IF, 1:200) Anti-eMYHC (Sigma-Aldrich, MFCD00162703; IF, 1:200) Anti-Laminin (Proteintech, 23498-1-AP; IF, 1:200) Anti-Pax7 (Sigma-Aldrich, 12352203; IF, 1:200) Anti-Fast Myosin (Servicebio, GB112130-100; IF, 1:200) Anti-Slow Myosin (Servicebio, GB112131-100; IF, 1:200) Primary antibodies for western blot (WB): Anti-CaMK (Servicebio, GB11812-100; WB, 1:1000) Anti-phospho-CaMK (Thermo Fisher Scientific, PA5-37833; WB, 1:1000) Anti-AMPK (Proteintech, 10929-2-AP; WB, 1:1000) Anti-phospho-AMPK (Proteintech, 86794-1-RR; WB, 1:1000) Anti-HIF-1α (Servicebio, GB114936-100; WB, 1:1000) Anti-GAPDH (Servicebio, GB11002-100; WB, 1:5000) Secondary antibodies: Anti-mouse HRP (Abcam, ab47827; WB, 1:5000) Anti-rabbit HRP (Abcam, ab7090; WB, 1:5000) Anti-mouse Alexa Fluor 555 (CST, #4409; IF, 1:500) Anti-rabbit Alexa Fluor 555 (CST, #4413; IF, 1:500) Anti-mouse Alexa Fluor 488 (CST, #4408; IF, 1:500) Anti-rabbit Alexa Fluor 488 (CST, #4412; IF, 1:500) Secondary antibodies used: Anti-mouse HRP (Abcam, ab47827; WB, 1:5000), Anti-rabbit HRP (Abcam, ab7090; WB, 1:5000), etc. Where appropriate, antibodies were applied according to standard immunofluorescence or western blot protocols. Study design Fully implantable MEBIS fabrication The fully implantable MEBIS comprised a flexible M-PTNG and PCNC. The manufacturing of the M-PTNG primarily relied on the layer-by-layer assembly technique and incorporated a fully encapsulated design. To achieve high output performances of M-PTNG, the hierarchically porous PVDF/MXene film with remarkable dielectric properties was meticulously designed using a vapor-induced phase-separation process and further optimized to function as a triboelectric layer. Additionally, the regenerated silk nanofibers film, fabricated through an electrospinning technique, was utilized as the other triboelectric layer. Importantly, to ensure a stable electrical output of the M-PTNG in vivo, the Kapton film and PDMS were used successively as the first and second encapsulation layers. Fabrication of silk nanofibers film Regenerated silk fibroin films were produced from degummed Bombyx mori silk fibers78. Initially, the degummed silk was dissolved in a solution of calcium chloride (CaCl2) and formic acid (FA) at room temperature, resulting in an 8 wt % solution. Regenerated silk fibroin films were produced by pouring a silk-CaCl2-FA solution (8 wt %) into polystyrene Petri dishes. The dishes containing the silk solution were then placed in an airing chamber, facilitating rapid evaporation of formic acid within 24 h. Subsequently, the resulting silk films were immersed directly in water for 2 h to eliminate CaCl2 and residual formic acid, thereby inducing β-sheet structure. The regenerated silk films were dried at 60 °C for 4 h. Then, the silk fibers were dissolved in 98% formic acid to create a 12 wt % spinning solution. A small amount of poly (ethylene oxide) (PEO, ~0.24 wt %) was added to enhance fiber formation. Electrospinning was performed using an applied voltage of 15 kV and a solution flow rate of 1 mL/h, with the needle tip to metallic frame collector distance set at 15 cm. The thickness of the silk nanofibers film was easily adjustable by varying the spinning time. Finally, the obtained silk nanofibers films were air-dried and then placed in a vacuum dryer for 24 h. Preparation of Ti3C2Tx MXene nanosheets Ti3C2Tx MXene nanosheets were synthesized through an acidic etching process of Ti3AlC2 powders (MAX phases, particle size <38 µm) using a previously reported method79,80. Initially, 1 g of Ti3AlC2 powders was gradually introduced into an etchant solution consisting of 1.6 g of lithium fluoride (LiF) and 20 mL of 9 M hydrochloric acid (HCl). The mixture was continuously stirred at 50 °C for 30 h to ensure a complete reaction. Subsequently, the resulting product was washed with deionized (DI) water five times, with each washing step involving 5 min of centrifugation at 3500 rpm. As the pH of the mixture approached neutrality, the collected sediment was redispersed in DI water and subjected to sonication under an ice bath for 20 min to delaminate the clay-like MXene. The self-delaminated MXene was then centrifuged at 1500 rpm for 30 min to remove any unexfoliated MAX sediment, and the supernatant was carefully collected. The resulting dispersion was further centrifuged for 20 min at 4500 rpm, resulting in the observation and collection of a dark-green supernatant solution containing Ti3C2Tx MXene nanosheets. Finally, the obtained MXene nanosheets dispersion was sealed and stored at approximately 4 °C. Fabrication of hierarchically porous PVDF/MXene film The obtained MXene dispersion was collected using vacuum-assisted filtration and then dried under vacuum at 60 °C. Poly (vinylidene fluoride) (PVDF, 15 wt%) powder was dissolved in N, N-dimethylformamide (DMF) and acetone with a volume ratio of 3:2 at 60 °C with stirring. Subsequently, a suitable mass fraction of the as-synthesized MXene powders was added to the PVDF dispersion and sonicated for 30 min, which was followed by stirring overnight at room temperature. The obtained PVDF/MXene solution (2 mL) was dipped onto a silicon (Si) wafer substrate and then put into a spin-coater (KW-12A, Start, China) with a speed of 500 rpm min−1 in 10 s. The Si wafer coated with PVDF/MXene solution was exposed to a thermostatic chamber (DK-8D, Jinghong, China) at 65 °C and a relative humidity of ~70% for 30 min to generate a porous structure. After the vapor-induced phase-separation process, the composite films were removed from the Si wafers and immersed in an ethanol solution, followed by sonication for 10 min to remove any residual DMF and acetone solvent. Finally, the hierarchically porous PVDF/MXene films were obtained after a hot-pressing process at 60 °C. As-received PVDF/MXene films were stored in a vacuum for further operation. Designing of M-PTNG The as-fabricated M-PTNG contains triboelectric layers, electrodes, spacers, and encapsulation layers. In detail, a hierarchically porous PVDF/MXene film was used as the bottom triboelectric layer, and a conductive copper (Cu) film was attached to the PVDF/MXene film, serving as one of the electrodes. The silk nanofibers film and the Cu film were respectively utilized as the top triboelectric layer and the other electrode. The soft wire was fixed on the back of the two electrodes by applying the silver paste. Furthermore, spacers made from acrylic tape were incorporated into the two opposite triboelectric pairs to ensure highly effective contact and separation. The Kapton film was used as the first package layer. Next, PDMS (Sylgard 184 silicone rubber, Dow Corning) was mixed with the curing agent at a ratio of 10:1 and then spin-coated onto the Kapton film as the second package layer. Subsequently, it underwent a vacuum treatment and was solidified at 80 °C for 2 h to eliminate any voids and ensure the protection of M-PTNG from the surrounding environment. Fabrication of PCNC To fabricate PCNC, a reciprocating motion pattern of 80 mm was first designed using AutoCAD 2020. The pattern was then converted into G-code compatible with an EDW system using Autodesk ArtCAM 2018. A total of 10 g of polycaprolactone (PCL, Mn = 80,000) was dissolved in 30 mL of trifluoroethanol (TFE) in a 100 mL glass bottle equipped with a magnetic stir bar. The solution was stirred overnight at room temperature to ensure complete dissolution. Subsequently, 8 mL of the PCL solution was loaded into a 10 mL syringe, which was mounted onto the EDW system’s dispensing pump. A rotating tungsten carbide rod served as the receiving substrate to facilitate the formation of tubular structures. The printing parameters, including feeding rate, reciprocating speed, and an applied voltage of 2.8 kV, were systematically optimized. The distance between the extrusion nozzle and the rotating rod was maintained at approximately 2 mm. The rotation speed of the tungsten rod was adjusted to ensure stable fiber deposition and uniform conduit wall formation. Pre-tests were conducted to monitor the material consumption over time, and printing was terminated once the conduit reached a predefined outer diameter. The fabricated PCL conduits were subsequently placed in a vacuum desiccator to remove residual solvent. Conduits were cut into appropriate lengths as required by the experimental design, and washed repeatedly with absolute ethanol and ultrapure water to eliminate surface impurities, followed by drying and UV sterilization. To functionalize the conduits, a total of 20 mg of rGO was accurately weighed and dispersed in 20 mL of deionized water. The mixture was subjected to ultrasonication for 30 min to ensure the formation of a uniform and stable suspension. The dried PCL conduits were immersed in the rGO solution and placed on a shaker overnight to allow uniform surface coating. Subsequently, a mixture of pyrrole and hydrochloric acid was added, followed by 30 s of sonication to facilitate full permeation of the solution into the conduit matrix. Samples were pre-incubated at 4 °C for 1 h. For in situ polymerization of PPy, ferric chloride solution was introduced into the system, yielding final concentrations of 14 mM pyrrole, 14 mM HCl, and 38 mM FeCl3. The reaction was allowed to proceed under gentle shaking at 4 °C for 24 h to ensure uniform deposition of PPy onto the conduit surfaces. Finally, the conduits were washed thoroughly with absolute ethanol and ultrapure water, dried, and sterilized by ultraviolet irradiation prior to use. Performance testing of nerve conduits The physicochemical properties of the materials were characterized using various instruments, including SEM, a mechanical testing apparatus, a multimeter, an electrochemical workstation, and a four-point probe. SEM was used to observe the morphological characteristics of the conduits. For mechanical testing, the conduits were vertically fixed in a tensile machine (H5K-S Universal Testing Machine) for stretching experiments. Conductivity was measured using the four-point probe method (VSC) to calculate the electrical conductivity. Characterization of MEBIS SEM (Hitachi S-4800) and energy-dispersive X-ray spectroscopy were conducted to study the surface/cross-sectional morphologies and elemental dispersion of samples. The structural property of MXene nanosheets was characterized using TEM (JEM-2100F). The X-ray diffraction (XRD, Rigaku Ultima IV) patterns of the samples were measured by an X-ray diffractometer with Cu Kα radiation (λ = 1.54178 Å). The deconvoluted XRD spectra were used to calculate the phase fractions within the PVDF/MXene films from the intensities for the respective peaks following Eqs. (1)–(3): Here, Iα, Iβ, and Iγ represent the intensities of the peaks observed at 19.3°, 20.2°, and 18.1°, respectively. XPS (Thermo Scientific ESCALAB 250Xi) and Raman (Renishaw inVia) spectra analysis were employed to characterize the chemical composition and valence states of MXene and the interaction between MXene and PVDF. The dielectric properties of PVDF/MXene film were measured using an impedance analyzer (Agilent 4294A). A modal shaker (SA-JZ020) was applied as the external force (operating distance, 1 cm). Furthermore, the output voltage and current of the M-PTNG were measured using a Keithley 6514 electrometer system, and a force gauge (HZC-30A) was utilized to measure the pressure. The physicochemical and functional properties of the fabricated nerve conduits were comprehensively evaluated using SEM, mechanical testing, four-point probe conductivity measurement, and additional biocompatibility assays. Surface morphology The surface topography and microstructural features of the nerve conduits were examined using SEM (Hitachi SU8010, Japan) at an accelerating voltage of 5 kV after gold sputtering. This analysis enabled visualization of the fiber distribution, wall integrity, and overall morphology of the printed scaffolds. Mechanical testing To assess the tensile properties, nerve conduit samples were vertically mounted on a universal testing machine (Instron 5943, USA) and subjected to uniaxial stretching. The tests were performed under controlled environmental conditions: ambient temperature of 20 °C and relative humidity of 60%. The tensile rate was set at 10 mm/min. Stress–strain curves were recorded to determine mechanical parameters such as tensile strength and elongation at break. Electrical conductivity The electrical conductivity of the nerve conduits was measured using a standard four-point probe method (VSC-4, Suzhou Jingge Electronic, China). Conductivity was calculated based on the sheet resistance and sample dimensions using the appropriate geometric correction factors. Degradation assessment To evaluate the biodegradability of the conduits, samples were immersed in phosphate-buffered saline (PBS, pH 7.4) and incubated under physiological conditions. At predetermined time intervals, samples were retrieved, rinsed, dried, and weighed to calculate the percentage of mass loss, which was used to quantify degradation kinetics over time. Hemolysis assay The hemocompatibility of the conduits was assessed via hemolysis testing. Briefly, samples of the PCL and PCNC materials were placed in centrifuge tubes and equilibrated in DPBS for 15 min. In parallel, fresh rat blood was diluted by mixing 200 μL of whole blood with 250 μL of DPBS. Three experimental groups were established in a 6-well plate: ultrapure water as the positive control (P.C.), DPBS as the negative control (N.C.), and the test groups (PCL and PCNC). Diluted blood was added to each well, ensuring full contact with the test materials. The plates were incubated at 37 °C for 1 h, followed by centrifugation at 5000 rpm for 3 min at room temperature. Subsequently, 100 μL of supernatant from each group was transferred to a new plate, and optical density (OD) values were measured at 545 nm using a microplate reader (BioTek Synergy HTX, USA). The hemolysis rate (%) was calculated using the following equation: Assembly of the MEBIS To facilitate subsequent in vivo experiments, a customized MEBIS was assembled with tailored structural and dimensional parameters. Massage actuators of appropriate size and elasticity were designed to accommodate the anatomical features of the animal model and ensure efficient mechanical stimulation. The system was constructed by integrating a programmable reciprocating actuator, a calibrated pressure sensor, a mechanical output monitor, a massage actuator, and a voltage detection unit. Components were interconnected following a pre-established schematic to ensure functional compatibility and signal integrity. After assembly, the system underwent preliminary validation to assess operational stability, including repeatability of motion, real-time feedback accuracy from the pressure sensor, and consistency of mechanical output. Electrical monitoring was conducted in parallel to verify the fidelity of stimulation signals delivered during simulated massage cycles. Biocompatibility and cell viability assays CCK-8 assay SCs were purchased from the Cell Bank of the Chinese Academy of Sciences, Shanghai, China (catalog number: AGNR 6). These SCs were used for all subsequent in vitro experiments. Cells were cultured for predetermined time points, after which the CCK-8 reagent was prepared following the manufacturer’s protocol. The reagent was added to each culture dish to ensure complete contact with the cells. The plates were returned to a CO2 incubator for a designated period to allow the reagent to react fully. Absorbance was then measured at the specified wavelength using a microplate reader. Optical density (OD) values at different time points were recorded, and cell viability/proliferation was calculated based on these values, with statistical analysis performed thereafter. LIVE/DEAD staining Following the predetermined culture period, the medium was aspirated, and cells were rinsed once with PBS. A detection solution containing Calcein AM and propidium iodide (PI) was added according to the supplier’s instructions. The cells were incubated at 37 °C in the dark for 30 min. Post-incubation, fluorescence images were acquired using a fluorescence microscope (Calcein AM: green fluorescence, Ex/Em = 494/517 nm; PI: red fluorescence, Ex/Em = 535/617 nm). Cell viability and death rates were quantified based on the fluorescent signals. Cytoskeleton staining Pieces of polydimethylsiloxane (PDMS) from the nanogenerator packaging were cut into appropriate shapes and disinfected by immersion in 75% ethanol for 30 min. After repeated PBS washes and air-drying, the samples were UV-sterilized (both sides, 30 min each). SCs were then seeded onto the PDMS surfaces and incubated for 3 days. The cells were fixed and permeabilized before applying Actin-Tracker Red staining solution, followed by a 30-min incubation at room temperature in the dark. After three washes with Tween/PBS, the cells were counterstained with 0.2 mg/mL DAPI at 4 °C for 1 h, mounted with glycerol, and subsequently examined using laser confocal microscopy. Testing and simulation of MEBIS In vivo testing Following nanogenerator implantation, anesthetized rats were secured in a custom-designed frame in the supine position with the gastrocnemius muscle exposed. The assembled massage actuator was adjusted and connected to both mechanical and electrical monitoring devices. The applied massage force was incrementally adjusted while the corresponding voltage outputs were recorded. During the gradual application of mechanical stress by the massage actuator, real-time ultrasound imaging captured the resulting deformation of the gastrocnemius muscle. Simulation studies COMSOL Multiphysics software was employed to simulate the dynamic interaction between the massage actuator, the nanogenerator, and muscle tissues. The simulation assessed both the mechanical stress distribution and the resulting electric field formation surrounding the nerve conduit. ES and its impact on cellular behavior ES protocol SCs were seeded on the surface of the conduit materials. The experimental group was subjected to daily ES using a nerve stimulation system, while a parallel non-stimulated group served as the control. The system delivered pulsed electrical signals through a massage actuator that directly applied mechanical forces to a nanogenerator, with the generated current transmitted via connecting wires to the conduit. Immunofluorescence analysis post-stimulation After the intervention period, cells were fixed in 4% paraformaldehyde for 10 min and washed twice with PBS. Permeabilization was achieved with 0.1% Triton X-100 for 10 min, followed by two additional PBS washes. Fluorescently labeled primary antibodies(see Antibodies section) against S100 and GFAP were applied, and the samples were incubated overnight at 4 °C. After washing, secondary antibodies were added and incubated at 37 °C for 1 h. Nuclei were counterstained with 1 μg/mL DAPI, and the expression of neural markers was visualized using confocal microscopy. Quantification of neurotrophic factor secretion by ELISA For the analysis of secreted factors (VEGF, NGF, BDNF, and NT-3), ELISA plates were first coated with diluted capture antibodies (adjusted to 5 μg/mL in coating buffer, 0.1 mL per well) and incubated at 4 °C overnight. The following day, wells were washed thrice. Culture supernatants collected from the 3-day proliferation samples were diluted appropriately and added (0.1 mL per well), followed by incubation and subsequent washing. Enzyme-linked secondary antibodies were then applied, and after further incubation and washing steps, substrate solution was added for color development. The reaction was terminated, and absorbance was measured at 450 nm with a microplate reader. Concentrations of VEGF, NGF, BDNF, and NT-3 were determined by comparison with a standard curve. Calcium ion imaging To evaluate calcium channel activation following defined treatment durations, cells were seeded in 24-well plates containing conductive conduit materials. ES was applied according to the experimental grouping, after which the culture medium was removed, and the cells were gently rinsed with PBS. A volume of 250 μL of Fluo-4 AM staining solution was added to each well, followed by incubation at 37 °C in the dark for 30 min. Fluorescence imaging (Ex/Em = 490/525 nm) was performed to evaluate intracellular calcium levels. Ethics All animal experiments were conducted in accordance with institutional guidelines for the care and use of laboratory animals and were approved by the Animal Ethics Committee of Shanghai Tenth People’s Hospital (Approval No. SY2021-7035). Male Sprague–Dawley rats (Crl:CD®(SD), Charles River, China), approximately 200 g and 12 weeks old, were obtained from Shanghai Southern Model Biotechnology Development Co., Ltd. (Shanghai, China). Male Bama miniature pigs (Bama Xiang pigs), 3 months old, were purchased from the Experimental Animal Center of Southern Medical University, China. All animals were housed under standard controlled conditions (21 ± 3 °C, 55 ± 5% humidity, 12-h light/dark cycle) with free access to food and water. SD rats surgical procedure For rat experiments, animals were randomly divided into four groups: (i) PI group, (ii) AUTO, (iii) NEPI group, and (iv) MEBIS group. Under intraperitoneal anesthesia with 3% pentobarbital sodium (0.9 mL/kg), a skin and muscle incision was made to expose the sciatic nerve, which was transected to create a 12-mm defect. The defect was bridged using the designated scaffolds or an autologous nerve (reversed 180° anastomosis), and nanogenerator components were implanted on the ipsilateral gastrocnemius muscle. Postoperatively, antibiotics were administered intraperitoneally to prevent infection, and wound healing was closely monitored. Postoperative rehabilitation and imaging Beginning two weeks post-surgery, rats underwent routine massage stimulation (30 min/day, cycling 10 min of massage with 5 min rest, five times per week). Computed tomography (CT) imaging was performed at 1 week and 3 months post-surgery to assess device stability. Anesthetized rats were scanned in the prone position using appropriate CT parameters, and the DICOM data were processed with RadiAnt DICOM Viewer for three-dimensional reconstruction. Functional assessment of hindlimb recovery To evaluate the efficacy of MEBIS in promoting skeletal muscle recovery, multiple complementary functional assays were performed at defined post-injury time points. Toe-spread reflex Hindlimb muscle function was assessed using the toe-spread reflex, adapted from previous studies9. Rats were gently covered with a cloth and lifted by the tail to allow full extension and splaying of the toes. Reflex responses were scored as follows: 0 = no spreading, 1 = partial spreading, and 2 = full spreading (Fig. 6a). Assessments were conducted at 0, 4, 8, and 12 weeks post-surgery. In vivo muscle isometric force measurement Isometric contractile force of the TA muscle was measured in anesthetized rats using established protocols9. Animals were anesthetized with isoflurane, and the hindlimb was positioned with the knee fixed while the foot was attached to a footplate connected to a servomotor. Two monopolar electrodes were inserted subcutaneously into the medial and lateral sides of the TA muscle. Optimal muscle length for maximal force generation was determined by adjusting the footplate relative to the knee and applying single electrical pulses until force plateaued. Muscles were subsequently stimulated across a defined frequency range, with 2-min intervals between stimuli. Force data were recorded and analyzed using the 611 A Dynamic Muscle Analysis software. Maximal isometric force values were obtained for each animal, and animals were returned to their home cages after measurements. Motor function analysis To assess locomotor activity, rats were individually placed in a square open-field arena with sides measuring approximately 1 meter. Animals were allowed to acclimate until spontaneous locomotion stabilized, after which 10 min of video recordings were captured to monitor natural motor behavior. Recorded videos were subsequently analyzed using the Smart 3.0 behavioral tracking software to quantify parameters, including total distance traveled and average movement speed. Gait analysis and sciatic functional index (SFI) At 4, 8, and 12 weeks post-surgery, footprint analysis was performed to calculate the SFI using parameters such as toe spread (TS), print length (PL), and intermediate toe spread (IT) according to the formula: SFI = (−38.3 × (EPL − NPL) ÷ NPL) + (109.5 × (ETS − NTS) ÷ NTS) + (13.3 × (EIT − NIT) ÷ NIT) − 8.8, where values range from −100 (complete loss) to 0 (normal function). Electrophysiological measurements At 12 weeks, electrophysiological recordings were obtained following sciatic nerve exposure. Stimulation was applied at defined points, and parameters such as NCV and CMAP were recorded. Histological and ultrastructural analysis At 3 months post-surgery, regenerated nerve tissues were harvested. Morphological evaluation was performed using HE and TB staining, while TEM quantified axonal density and myelin thickness. Immunofluorescence staining was performed using antibodies against neural markers (S100, β-III-tubulin, MBP, and NF200) and vascular markers (CD31 and CD34) (see Antibodies section for details). Additionally, the wet weights of the triceps surae muscles from both the operated and contralateral sides were measured, and muscle fiber morphology was analyzed via HE staining. Fluoro-gold (FG) retrograde tracing To assess axonal regeneration and neuronal connectivity, retrograde tracing was performed using Fluoro-Gold (FG; Fluorochrome, Denver, CO, USA). At 12 weeks post-sciatic nerve injury and MEBIS treatment, animals were anesthetized with isoflurane, and 10 μL of 4% FG solution was injected into the distal nerve stump of the injured limb using a Hamilton syringe. Care was taken to avoid leakage and ensure even diffusion. After 5 days, animals were perfused with 4% paraformaldehyde (PFA), and the spinal cords and dorsal root ganglia (DRGs, L4–L5) were dissected, post-fixed overnight, and cryoprotected in 30% sucrose. Spinal cord and DRG samples were embedded in OCT and sectioned at 10 μm thickness. Sections were mounted on slides. FG-labeled neurons were visualized using a fluorescence microscope (Zeiss Axio Observer or equivalent). Images were captured under consistent exposure settings across all groups. Scale bars were set at 500 μm for spinal cords and 200 μm for DRGs. The number of FG-positive motoneurons and DRG sensory neurons was counted in at least three representative sections per animal. Counts were performed in a blinded manner using ImageJ. Neurotrophic factor quantification in distal nerve stumps To quantify in vivo neurotrophic factor production and release within the regenerating nerve microenvironment, distal sciatic nerve stumps were harvested from rats at 4, 8, and 12 weeks after surgery. Excised tissues were immediately snap-frozen in liquid nitrogen and homogenized in ice-cold lysis buffer containing protease inhibitors. Tissue homogenates were centrifuged, and the supernatants were collected for protein analysis. Total protein concentrations were determined using a bicinchoninic acid (BCA) assay to ensure equal loading across samples. The protein levels of vascular endothelial growth factor (VEGF), nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-3 (NT-3) were quantified using commercially available enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturers’ instructions. Absorbance was measured using a microplate reader, and concentrations were calculated based on standard curves generated for each analyte. Neurotrophic factor levels were normalized to total protein content and expressed relative to the AUTO group at each corresponding time point. Heatmaps were generated using GraphPad Prism, with color intensity indicating fold changes in protein abundance. Molecular analyses RNA sequencing and differential expression analysis Total RNA was isolated from sciatic nerve tissues using TRIzol reagent (Invitrogen, USA) and quantified with a NanoDrop 2000 (Thermo Fisher Scientific, USA). cDNA libraries were constructed and sequenced on an Illumina HiSeq X10 platform (Illumina, USA). Gene expression was quantified in fragments per kilobase of transcript per million mapped reads (FPKM). Differential expression analysis was performed using the DESeq R package, with significant changes defined by FDR < 0.05 and a threshold for |log₂ fold change|. Enrichment analyses were carried out using the clusterProfiler package in R. Tissue validation by Western blot and qPCR Collected tissue samples were lysed on ice using a lysis buffer (Thermo Fisher Scientific, USA). Protein concentrations were determined using a BCA assay (Thermo Fisher Scientific, USA). Proteins were denatured by mixing with SDS loading buffer and heating at 100 °C for 7 min, then separated by SDS-PAGE and transferred onto nitrocellulose membranes (Bio-Rad). Membranes were blocked with 3% BSA for 1 h at room temperature, incubated overnight at 4 °C with primary antibodies, washed with TBST (three washes, 5 min each), and incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Bands were visualized using an ECL detection system, and band intensities were quantified using ImageJ, normalizing target proteins to internal controls. Uncropped and unprocessed scans of the original western blots, including visible molecular weight markers, have been provided in the Source data file (uploaded in Excel format). Total RNA was extracted using TRIzol reagent, and its purity was assessed by the A260/280 ratio using a NanoDrop spectrophotometer (Thermo Fisher Scientific, USA). RNA integrity was verified via 1% agarose gel electrophoresis. cDNA was synthesized from 1 μg of total RNA using the PrimeScript™ RT reagent kit (Takara, Japan). qPCR was performed using SYBR Green Premix Ex Taq™ II (Takara, Japan) on a CFX96 Real-Time PCR Detection System (Bio-Rad, USA) with the following cycling parameters: initial denaturation at 95 °C for 30 s, followed by 40 cycles at 95 °C for 5 s and 60 °C for 30 s. A melting curve analysis from 65 to 95 °C confirmed amplification specificity. GAPDH was used as the internal reference gene, and relative gene expression was calculated using the 2−ΔΔCt method. Each reaction was performed in technical triplicates and repeated at least three times independently. Primer sequences used for qPCR are provided in Supplementary Table 3. In vivo biosafety evaluation Biosafety was assessed by immunohistochemical staining of tissues surrounding the implant for inflammatory markers (e.g., CD68 and TNF-α) (see Antibodies section for details) as well as histological evaluation (HE staining) of major organs, including the heart, liver, spleen, lung, kidney, and brain. Bama miniature pig evaluation Male Bama miniature pigs (approximately 20 kg) were randomly allocated into four groups: PI group, AUTO, NEPI group, and MEBIS group. Preoperatively, pigs were fasted for 8 h with water withheld for 3 h. A compound sedative (atropine, midazolam, and ketamine) was administered intramuscularly, followed by induction with propofol (2–4 mg/kg via the auricular vein) and endotracheal intubation. Anesthesia was maintained with isoflurane inhalation. A surgical incision was made to expose the sciatic nerve, which was transected to create a 30-mm defect. The defect was bridged using the designated scaffold or autologous nerve (reversed 180° anastomosis), and nanogenerator components were implanted on the ipsilateral gastrocnemius muscle. Postoperative care included wound management, thermoregulation, and prophylactic intraperitoneal antibiotic administration. Muscle status was monitored by ultrasound, gait was recorded, and tissues from the gastrocnemius and regenerated nerve were harvested for histological and TEM analyses. Statistics and reproducibility All experiments were independently repeated at least three times with similar results unless otherwise stated. Representative images are shown where appropriate. The number of independent experiments (n) for each assay is indicated in the corresponding figure legends. All data collected in this work are expressed as means ± SD. One-way analysis of variance (ANOVA) or two-way ANOVA with Tukey’s multiple comparisons test with Tukey’s multiple comparisons test (two-sided) was performed to identify significant differences between groups. To compare the means of two independent groups, an unpaired Student’s t-test was performed. Differences were considered significant at *p < 0.05 and **p < 0.01 or were considered highly significant at ***p < 0.001. All statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, USA) and ImageJ 1.53 (NIH, USA) for image analysis. For more details of the fabrication, please see Supplementary Materials and Methods. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Reporting Summary Transparent Peer Review file Source data Source data Acknowledgements The authors would like to thank Yao Miao, Fan Zhang, and Zhong Wu for their valuable assistance and support during the course of this study. ChatGPT (OpenAI) was used for language polishing of the manuscript. Figures were created using BioRender.com. The authors take full responsibility for the content of this work. Author contributions Conceptualization: L.N., Y.Q., W.C. and F.C.; methodology: J. L, F.C., J. F, and W.C.; investigation: L.N., J.F., X.J., J.W., F.J, S.L. and K. S.; visualization: F. W, Z.Z., J.X. and T.W.; supervision: J. L, F.C., W.C. and Y. Q.; writing—original draft: L.N., J.F., and W.C.; writing—review and editing: J. F, L.N., P.S., Y. Q, W.C. and F.C. Peer review Peer review information Nature Communications thanks Rolf Heumann, Banafsheh Dolatyar, and the other, 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 National key R&D Program of China (No. 2021YFC2400801 to Y.Q.), National Natural Science Foundation of China (52202108 to W.T.C. and 82372409 to Y.Q.), Natural Science Foundation of Shanghai’s “Science and Technology Innovation Action Plan” (No. 23ZR1449000 to J.J.L.), The Key Cultivation Project of “Medicine +X” Interdisciplinary Research of Tongji University (No. 2025-0553-ZD-07 to J.J.L.), Sino-German Mobility Programme (Grant No. M-0699 to Y.Q.), Cultivation grant for clinical and basic integration research of Shanghai Tenth People’s Hospital (SYYYRH2025047 to L.P.N.). Data availability The raw RNA sequencing data generated in this study have been deposited in the NCBI BioProject database under accession number PRJNA1429883 and are available via the Sequence Read Archive (SRA). The data will be publicly available upon publication. Processed data supporting the findings of this study are provided in the Supplementary Information. Source data underlying the figures are provided with this paper. All other data supporting the findings of this study are available within the paper and its Supplementary Information files. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Liping Nan, Wentao Cao, Jiaqi Fang, Pengzhi Shi. These authors jointly supervised this work: Wentao Cao, Feng Chen, Junjian Liu, Yun Qian Contributor Information Wentao Cao, Email: wt_cao@fudan.edu.cn. Feng Chen, Email: chen_feng@fudan.edu.cn. Junjian Liu, Email: jjliu@tongji.edu.cn. Yun Qian, Email: sakio@sjtu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73535-w. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Reporting Summary Transparent Peer Review file Source data Data Availability Statement The raw RNA sequencing data generated in this study have been deposited in the NCBI BioProject database under accession number PRJNA1429883 and are available via the Sequence Read Archive (SRA). The data will be publicly available upon publication. Processed data supporting the findings of this study are provided in the Supplementary Information. Source data underlying the figures are provided with this paper. All other data supporting the findings of this study are available within the paper and its Supplementary Information files. Source data are provided with this paper.