Wireless flexible patch based on ultrasound-responsive SenA@ZIF-8 release for on-demand medication of periodontitis Authors: Bo Peng, Jing Yang, Rui Lin, Yuling Han, Hongning Wang, Runqi Zhou, Tong Yan, Shilin Liu, Hanwei Jiang, Yunshan Li, Tianshu Zhang, Yuhao Peng, Xinyu Xue, Huawei Shen, Jinlin Song, Ximu Zhang Published: 2026 ABSTRACT AbstractPeriodontitis is a prevalent chronic inflammatory disease that causes progressive periodontal tissue loss and systemic complications. However, conventional therapies often remove biofilms without fully resolving inflammation or regenerating tissue. Here, we report a flexible wireless ultrasonic patch loaded with ZIF-8 nanoparticles containing the anti-inflammatory antioxidant senkyunolide A (SenA@ZIF-8) for on-demand drug release. A piezoelectric transducer converts 1 MHz ultrasound into ≈ 10 V, 9 mA electrical output, heating a tridecanoic acid phase-change matrix to ≈ 43 °C to trigger SenA@ZIF-8 release. In vitro, SenA@ZIF-8 was biocompatible and outperformed free SenA in protecting periodontal ligament stem cells from H₂O₂-induced oxidative damage: it reduced intracellular ROS levels and restored the activity of antioxidant enzymes. In macrophages, ZIF-8 loading enhanced intracellular SenA uptake and markedly suppressed LPS-induced proinflammatory mediator expression while promoting M2 marker expression. In a rabbit model of ligature-induced periodontitis, repeated ultrasonic-triggered SenA@ZIF-8 delivery markedly reduced periodontal inflammation. Micro‑computed tomography (Micro-CT) and histology showed that the device-treated sites exhibited the greatest alveolar bone regeneration and collagen deposition, accompanied by minimal inflammation. This innovative MOF-based ultrasonic patch can relieve oxidative stress and inflammation through minimally invasive, precise, and on-demand drug delivery, thereby promoting periodontal regeneration. This approach has significant translational potential as an adjunct therapy to improve periodontal treatment outcomes and patient compliance.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12951-026-04609-6.Keywords: Periodontitis, Flexible patch, Ultrasonic power supply, ZIF-8, Senkyunolide A, Reactive oxygen species Introduction Periodontitis is a chronic inflammatory disease characterized by the progressive destruction of the periodontal soft tissue and alveolar bone. According to the Fourth National Oral Health Survey, nearly 90% of adults in China are affected to varying degrees, with more than 30% presenting with severe disease [1, 2]. Globally, periodontitis affects more than 700 million people and remains the leading cause of tooth loss in adults. In addition, it is a recognized risk factor for systemic disorders, including diabetes, atherosclerosis, and adverse pregnancy outcomes [3]. Despite extensive clinical efforts, effective prevention and control of disease progression remain major challenges. The pathogenesis of periodontitis is driven by gram-negative anaerobic bacteria and the host immune response. Bacterial colonization and plaque accumulation lead to the recruitment of neutrophils [4], which eliminate pathogens through phagocytosis and reactive oxygen species (ROS) generation. While ROS are essential for antimicrobial defense, their excessive production overwhelms the antioxidant capacity, induces oxidative stress, activates proinflammatory signaling, and promotes osteoclastogenesis, thereby accelerating alveolar bone resorption. ROS also directly damage the extracellular matrix, contributing to attachment loss [5]. Endogenous antioxidant systems, including superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase, are mobilized to maintain redox homeostasis. However, once this balance is disrupted, sustained oxidative stress exacerbates inflammation and impairs tissue regeneration. Therefore, controlling oxidative stress in the periodontal microenvironment after basic therapy is critical for periodontal repair [5]. Current treatments, such as scaling and root planing combined with local pharmacotherapy, effectively remove biofilms but often fail to sufficiently suppress inflammation or induce robust tissue regeneration [6]. Oxidative stress contributes to disease progression by inducing cellular dysfunction and amplifying inflammatory cascades [7–9]. Importantly, oxidative stress markers correlate positively with disease severity [7], highlighting ROS regulation as a promising therapeutic target. Accordingly, the development of ROS-scavenging materials capable of modulating the periodontal microenvironment is of substantial clinical interest. Based on the characteristics of periodontitis, we sought a single, chemically defined natural small molecule capable of simultaneously modulating inflammation and oxidative stress while remaining compatible with precise local delivery. Senkyunolide A (SenA) is a low-molecular-weight phthalide (C12H16O2, 192.25 Da) with limited polarity, moderate lipophilicity, and few rotatable bonds, consistent with a compact small-molecule scaffold that is favorable for reproducible quantification and efficient accommodation within porous carriers. Beyond its suitability for formulation, SenA is supported by mechanistic evidence that aligns with the inflammation–redox–tissue-repair axis. SenA has been reported to attenuate oxidative stress-associated injury by upregulating the SIRT6/Nrf2 pathway [10] and suppressing NLRP3-associated inflammatory programs, including the inhibition of inflammatory cytokine release, oxidative stress, and NLRP3-mediated ferroptosis in OGD/R-injured cells [11], as well as the inhibition of osteoarthritis progression via NLRP3 signaling [12]. In addition, SenA has been reported to disrupt the TRAF6–HDAC3 interaction and epigenetically suppress c‑MYC transcriptional programs [13]. Therefore, the choice of SenA provides a mechanistically coherent basis for coupling anti-inflammatory/antioxidant regulation with enhanced regenerative potential through local delivery. In parallel, zeolitic imidazolate framework-8 (ZIF-8), a Zn²⁺-based metal–organic framework, offers high stability, high pH responsiveness, and excellent biocompatibility [14–16]. ZIF-8 enables high drug loading and sustained release, while the released Zn²⁺ promotes osteogenesis and provides antibacterial activity [17, 18]. These features make ZIF-8 an attractive carrier for controlled SenA delivery in inflammatory periodontal niches [19]. Although scaling and root planing can increase therapeutic efficacy when combined with medication, the success of the entire treatment process depends not only on the properties of the drug but also on the delivery system and route of administration [20]. How to achieve minimally invasive, convenient, and precise local drug delivery after basic treatment remains one of the challenges requiring optimization in periodontitis management. In recent years, microdevices have been widely applied across various medical fields, such as controlled drug delivery platforms for treating acute conditions and chronic diseases [21]. Examples include local sustained drug delivery after cancer surgery [22], microcapsule-based oral vaccine delivery [23], and microcapsule-based antibiotic delivery for antimicrobial therapy [24]. In the field of oral medicine, flexible electronic devices are being increasingly utilized to advance preventive dentistry, oral medicine, and craniofacial disease management, as well as to assess corresponding prognoses [25]. Therefore, how to utilize the local drug delivery function of microdevices to ensure accurate drug delivery after the basic treatment of periodontitis and maintain stable effective drug concentration levels in the periodontal pocket [26] while minimizing administration trauma to facilitate epithelial regeneration [27] and increasing the convenience for patients to improve adherence and reduce the workload of medical staff [28, 29] have become clinical problems that need to be solved. Here, we report a flexible, wireless, ultrasound-responsive drug delivery system for targeted periodontal therapy. Tridecanoic acid is employed as a phase-change matrix to encapsulate SenA@ZIF-8 (ZS) particles and prevent premature release at physiological temperature. Upon external ultrasound stimulation, localized heating (~43 °C) melts the tridecanoic acid and triggers on-demand release directly into the periodontal pocket. This self-sustaining system enables precise spatiotemporal control of drug delivery, minimizes tissue trauma, and increases patient convenience. Collectively, this strategy integrates MOF-based drug delivery with flexible electronics to achieve minimally invasive, controllable, and regenerative therapy for periodontitis (Fig. 1). Schematic diagram. An ultrasonically driven periodontal pocket drug delivery device features three drug reservoirs per tooth. When medication is needed, ultrasound energy heats and releases the therapeutic agent SenA@ZIF-8, enabling precise, minimally invasive self-administration at home. The released drug reduces ROS levels, promotes osteogenesis, and exerts anti-inflammatory effects, thereby treating periodontitis Materials and methods Heater preparation First, the polyimide film substrate was sequentially cleaned with acetone, isopropanol, and deionized water, followed by drying with nitrogen. Next, the substrate was spin coated with the positive photoresist S1813 at 4000 rpm for 30 s to form an approximately 1.5 μm thick layer, which was then soft-baked at 90 °C for 1 min. Subsequently, ultraviolet lithography was performed using a mask with an exposure dose of 100 mJ/cm² for 5 s to define the serpentine electrode pattern. The exposed substrate was developed in MF-319 developer for 30 s, rinsed with deionized water, and dried with nitrogen. The substrate was then hard-baked at 120 °C for 2 min to increase the stability of the photoresist. Afterward, a 40 nm-thick chromium adhesion layer and a 200 nm-thick copper conductive layer were sequentially deposited via electron beam evaporation. Finally, ultrasonic lift-off was performed in acetone to remove the excess metal and residual photoresist, resulting in a well-defined serpentine electrode pattern on the polyimide film. Manufacture of the wireless ultrasonic patch First, we generated a flexible circuit on a polyimide substrate as the basic structure. Using photolithography and electron beam evaporation methods, we directly patterned and deposited serpentine heating electrodes on the substrate surface to achieve a precise shape and strong interfacial adhesion force (hence, an accurate electrode geometry and high interfacial adhesion were realized). After we prepared the electrode, the electronic components and integrated circuits (ICs) were soldered onto the flexible board to complete the circuit assembly. Polydimethylsiloxane (PDMS) was coated on the whole circuit through a mold-based casting process to increase the biocompatibility and provide a protective coating; it formed a box-shaped structure that acted as a drug reservoir. Ultrasound propagation simulation Finite element analysis (COMSOL) was used to simulate the spread of ultrasonic waves inside human soft tissues. Therefore, the model was constructed through the application of the solid mechanics and pressure acoustics modules to characterize ultrasound attenuation and directivity. All tissue parameters were acquired from material data sheets that the supplier had provided. Hence, the input conditions (ultrasound trigger signal: 43 V; propagation distance: 3 cm; 90° propagation in ultrasound gel) were fixed based on the experimental measurements (corresponding to Fig. 2C and D). Preliminary in vivo study and the working state of the ultrasonic wireless control system. (A–C) Effects of single- and multiple-dose intrapocket administration on the systemic antioxidant capacity, including the serum total antioxidant capacity, CAT activity, and SOD activity. (D) Surface SEM image of PZT. (E, F) When the AC ultrasound excitation is lower than 1 MHz, the output voltage and current from the PZT display a time delay, consistent frequency, and signal attenuation compared with the trigger signal. (G, H) Acoustic pressure distribution and attenuation of ultrasound in human soft tissue. (I) PZT signal rectification (AC→DC). (J–L) Rectified voltage vs. transducer-patch distance, ultrasound power, and incidence angle. (M) Voltage storage in 470/1000 μF capacitors. (N–P) Heating electrode (optical image), temperature change (curves), and infrared imaging (heating/cooling). (Q–T) Electrode heat-related properties (heating and cooling, maximum temperature, temperature fluctuation, response speed) under conditions with or without PDMS encapsulation and an assessment of device performance Synthesis of senA@ZIF-8 An equal volume of 0.6 mol/L methylimidazole solution in methanol was added dropwise into a 0.1 mol/L Zn(NO₃)₂ solution in N, N-dimethylformamide. The mixture was stirred at 380 rpm for 10 min at room temperature and then centrifuged at 14,000 rpm for 10 min. The solid precipitate was washed twice with anhydrous ethanol and then reconstituted with anhydrous ethanol to prepare a ZIF-8 solution. A SenA (62006-39-7; MACLIN, China) solution prepared in anhydrous ethanol was added to the ZIF-8 solution. The samples were incubated at room temperature in a dark environment for 14 h. The samples were centrifuged at 14,000 rpm, the supernatant was removed from the precipitate, and the ethanol solution was stored for future use. Representative photographs of the as-prepared ZS dispersion were taken with a camera (Nikon, Japan). Drug loading capacity and encapsulation efficiency of SenA@ZIF-8 After SenA loading, the suspension was centrifuged to separate the SenA@ZIF-8 particles from the supernatant containing the unloaded SenA. The supernatant was collected, and the concentration of free SenA was quantified by UV - Vis spectroscopy at 273 nm using a standard calibration curve prepared from SenA standards in the same solvent/matrix as the samples. The calibration curve showed excellent linearity, where y is the absorbance (OD = 273 nm) and x is the SenA concentration (µg/mL). The mass of free SenA in the supernatant was calculated as m = C × V, where C is the SenA concentration determined from the calibration curve and V is the total volume of the supernatant. The mass of the loaded/encapsulated SenA was calculated by mass balance: m (loaded) = m (input) – m (free). Loading capacity (LC, wt%) and encapsulation efficiency (EE, %) were calculated using the following standard definitions: LC (wt%) = m (loaded)/{m (ZIF-8) + m (loaded)} × 100%, and EE (%) = m (loaded)/m (input) × 100%. Isolation and culture of human periodontal ligament stem cells (hPDLSCs) Research related to human PDLSC donors was conducted according to ethical guidelines from the World Medical Association, and the Ethics Committee of Chongqing Medical University Affiliated Stomatological Hospital 2025 Ethics (076) approved this study. hPDLSCs were obtained from the periodontal ligament tissue of premolars that were planned to be removed during orthodontic extraction operations [30, 31]. All the donors were between 12 and 30 years old and provided written informed consent. The cells were cultivated in α-MEM (Gibco, USA) supplemented with 10% fetal bovine serum (Sigma, USA), 1% penicillin and streptomycin (Biosharp, China), and 1% GlutaMAX (Gibco, USA) at 37 °C in a humidified incubator with 5% CO₂. The Supporting Information provides related details. Toxic reactions, cell proliferation, half-maximal inhibitory concentration (IC₅₀), and cell recovery experiments using hPDLSCs The 96-well plates were seeded with 5,000 cells per well. hPDLSCs were seeded and cultured for one night; thereafter, except for the control group, the other groups were treated with different concentrations of ZIF-8, SenA, and SenA@ZIF-8 and cultured continuously for 48 h to detect toxic reactions. For the cell proliferation experiment, hPDLSCs were seeded and cultured overnight. Then, except for the control group, the cells in the other groups were cultured for 1, 3, or 5 days with either 40 µg/ml or 60 µg/ml ZIF-8, SenA, or SenA@ZIF-8. For the determination of the half-maximal inhibitory concentration, hPDLSCs were seeded and cultured overnight, after which different concentrations of H₂O₂ (Aladdin, 0.25 mM, China) were added to groups other than the control group, followed by 3 h of stimulation. For the cell recovery experiment, hPDLSCs were seeded and cultured overnight; then, 40 µg/ml or 60 µg/ml ZIF-8 or SenA@ZIF-8 was added to the control group for overnight coculture, followed by the addition of an equivalent concentration of H₂O₂ for 3 h of stimulation. Cell viability was measured quantitatively using the Cell Counting Kit-8 (CCK-8) assay (UElandy, China). More detailed information is provided in the Supporting Information. In vitro antioxidant test The ability of the cells to clear ROS was evaluated using an active oxygen test kit (2’,7’-dichlorofluorescein diacetate, DCFH-DA; S0033M, Beyotime, China; excitation wavelength, 488 nm; emission wavelength, 525 nm). After hPDLSCs were seeded (1 × 10⁴ cells/well, 48-well plate) and cultured overnight, 60 µg/ml ZIF-8, SenA, or SenA@ZIF-8 was added and incubated with the cells overnight. Afterward, the same concentration of H₂O₂ was added and incubated with the cells for 3 h before the cells were stained with DCFH-DA and Hoechst 33,342 (Beyotime, China; excitation wavelength, 346 nm; emission wavelength, 460 nm). ROS levels were measured with an inverted fluorescence microscope (Zeiss, Germany). Changes in the mitochondrial membrane potential were detected after the same culture conditions using a mitochondrial membrane potential detection kit (TMRE; C2001S, Beyotime, China; excitation wavelength, 550 nm; emission wavelength, 575 nm), a mitochondrial green fluorescence staining kit (Mito-Tracker Green; C1996M, Beyotime, China; excitation wavelength, 490 nm; emission wavelength, 516 nm), and Hoechst 33,342. Images were captured and quantified using ImageJ software. The average ROS level per cell was calculated as the total green fluorescence intensity/number of blue nuclei, and the average mitochondrial membrane potential activity per cell was calculated as the total red fluorescence intensity/total green fluorescence intensity. Therefore, the results were compared in a corresponding manner. hPDLSCs (7 × 10⁴ cells/well, 12-well plate) were seeded and cultured overnight before the intracellular levels of oxidative stress-related mRNAs were detected. Then, 60 µg/ml ZIF-8, SenA, and SenA@ZIF-8 were added for overnight coculture. After the cells were stimulated with the same reagents but a higher concentration of H₂O₂ for 2 h, RNA was extracted and purified with an RNA extraction kit (RNeasy™ Animal RNA Isolation Kit with Spin Column, R0027, Beyotime Tian, China). RT‒qPCR was used to analyze human catalase (hCAT), human glutathione peroxidase 1 (hGPX1), human superoxide dismutase 1 (hSOD-1), and heme oxygenase-1 (HO-1) expression in hPDLSCs. Additional details are provided in the Supporting Information. The primers used to determine the expression of the target genes are listed in Table S1. Osteogenic differentiation of hPDLSCs For the assessment of ALP activity, hPDLSCs (3 × 10⁴ cells/well, 24-well plate) were plated and cultivated for one night. Afterwards, except for the control group, the cells were cocultured for one night with 60 µg/ml ZIF-8, SenA, or SenA@ZIF-8. The cells were stimulated with the same volume of a lower concentration of H₂O₂ for 3 h, after which the culture medium was replaced with medium containing 10 mM β-glycerol phosphate sodium (Sangon, China), 0.05 mM L-ascorbic acid (Sigma, USA), and 10 nM dexamethasone (Aladdin, China). After 7 days of osteogenic induction, ALP staining was performed using a BCIP/NBT ALP color development kit (Beyotime, China). Moreover, the concentration of ALP was measured with an alkaline phosphatase assay kit (Beyotime, China) by detecting the absorbance at 405 nm. Based on the defined enzymatic activity, the alkaline phosphatase activity in the sample groups was calculated. The ARS staining process followed the same procedure. After 21 days of osteogenic induction, ARS staining was performed with a 1% alizarin red solution (ARS, Sigma, USA). After image acquisition, a quantitative analysis was performed using an Alizarin Red S Staining Quantification Kit (CPC, Solarbio, China), and the absorbance was recorded at 560 nm. The concentrations in the samples were calculated from the standard curve. More detailed information is provided in the Supporting Information. Culture of RAW264.7 cells RAW264.7 cells were purchased from the Chinese Academy of Sciences Cell Bank. The cells were cultured in high-glucose DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (Sigma, USA) and 1% GlutaMAX (Gibco, USA) at 37 °C in a humidified incubator containing 5% CO₂. Passaging and medium replacement strictly followed the protocols of medium rewarming and gentle pipetting actions to prevent the polarization of RAW264.7 cells. Toxicity and phagocytosis of RAW264.7 cells RAW264.7 cells (1 × 10⁴ cells/well, 96-well plate) were seeded in plates and cultured overnight before assessing toxicity. Different concentrations of SenA and SenA@ZIF-8 were added to wells that did not belong to the control group, after which the cells were incubated for 48 h. For the uptake test, RAW264.7 cells (1 × 10⁶ cells/well, 6-well plate) were seeded and incubated overnight. Equal amounts of SenA and SenA@ZIF-8 were added and cocultured for one night. After the supernatant was removed, the cells were collected, washed two times with PBS, and processed by sonication, after which the concentrations of SenA in the samples were determined. More detailed descriptions are provided in the Supporting Information. Inflammation resistance test in RAW264.7 cells RAW264.7 macrophages were seeded onto sterile glass coverslips in 24-well plates (5 × 104 cells/well) and allowed to adhere overnight. The cells were divided into the following groups: control, LPS, LPS + ZIF-8, LPS + SenA, and LPS + ZS. After 24 h of treatment, 0.5 µg/mL lipopolysaccharide (LPS; Sigma, USA) was added to the appropriate wells and incubated for 24 h to induce M1 polarization. The cells were fixed, permeabilized, and immunostained (Beyotime, China) with antibodies against the M1 marker CD86 (CD86 monoclonal antibody, MA1-10299, Thermo, USA) and the M2 marker CD206 (CD206 polyclonal antibody, PA5-101657, Thermo, USA), followed by fluorescent dye-conjugated secondary antibodies (goat anti-rabbit IgG H&L (AF488), Ex: 490 nm. Em: 525 nm; Rabbit Anti-Rat IgG H&L (TRITC), Ex: 552 nm. Em: 577 nm; Zen BioScience, China) and Hoechst 33,342 counterstaining. A fluorescence microscope was used to capture images of CD86 and CD206 expression. The fluorescence intensity per cell (normalized to the nuclear count) was quantified using ImageJ. We measured the levels of inflammation-related genes. RAW264.7 cells (1 × 10⁶ cells/well, 6-well plate) were seeded and cultured for one night. Afterward, 30 µg/ml ZIF-8, SenA, and SenA@ZIF-8 were added to the cultured cells and incubated for one night. Next, 0.5 µg/mL LPS was added and incubated with the cells for 6 h. RNA was subsequently extracted and purified using an RNA extraction kit. RT‒qPCR analysis was performed to measure the levels of the interleukin-6 (IL-6), interleukin-1β (IL-1β), inducible nitric oxide synthase (iNOS), tumor necrosis factor-α (TNF-α), arginase-1 (Arg-1), interleukin-10 (IL-10), and transforming growth factor-β (TGF-β) mRNAs in RAW264.7 cells. The primers used to determine the expression of the target genes are listed in Table S2. We also measured the levels of secreted inflammation-related proteins. RAW264.7 cells (1 × 10⁵ cells/well, 24-well plate) were seeded and cultured for one night. Afterward, 30 µg/ml ZIF-8, SenA, and SenA@ZIF-8 were added and incubated with the cells for one night. After 24 h of stimulation with 0.5 µg/mL lipopolysaccharide (LPS), the cell supernatants were collected. Nitric oxide levels were measured using a nitric oxide detection kit (S0021M, Beyotime, China), while IL-6 and TNF-α levels were measured using ELISA kits (Mouse IL-6 ELISA Kit, EK206; Mouse TNF-α ELISA Kit, EK282; MULTI, China). More detailed descriptions are provided in the Supporting Information. Transcriptome sequencing of RAW264.7 cells First, after we plated RAW264.7 cells at a density of 1 × 10⁶ cells per well in a 6-well plate and cultured them for one night, we added 30 µg/ml ZIF-8, SenA, and SenA@ZIF-8 and incubated the cells overnight. After this step was completed, 0.5 µg/mL lipopolysaccharide was added, and the cells were stimulated for a period of 4 to 6 h. RNA was subsequently extracted and purified using an RNA extraction kit, and the purified RNA was submitted to Meiji Company for transcriptome sequencing. The RNA-seq study was performed with three independent biological replicates per group (n = 3). Libraries were sequenced on an Illumina platform, yielding on average ~40 million high-quality paired-end reads (150 bp) per sample, with > 96% of bases having Q30 quality. Reads were aligned to the reference genome with an overall mapping rate > 93%. Differential gene expression was determined using DESeq2 with a significance threshold of adjusted p value (padj) < 0.05 and |log₂ FC | ≥ 1.0. As all the samples were prepared and sequenced together in a single batch, no additional batch effect correction was necessary. In vitro cytocompatibility of tridecanoic acid and testing the toxicity of the device A CCK-8 assay was performed to measure the viability of hPDLSCs and RAW264.7 cells and to evaluate whether tridecanoic acid was cytotoxic after ultrasound stimulation to induce melting/liquification. Cells were seeded in 96-well plates and allowed to adhere overnight. Tridecanoic acid was subsequently added to achieve final concentrations of 0, 10, 20, 40, 80, 160, 320, 640, and 1280 µg/mL. Two conditions were established in parallel for the experiment: −US (untriggered control group), where cells were coincubated with tridecanoic acid at 37 °C without ultrasound stimulation; and + US (triggered simulated group, US intensity: 2.5 W/cm², duty cycle: 100%, exposure time: 4 min, coupling medium: medical ultrasound gel, distance: 1 cm), where ultrasound stimulation was applied to the tridecanoic acid-treated cells in each well using a scheme identical to that used for device triggering. After the preset exposure incubation period, the cells were further cultured for 48 h, and cell viability was measured quantitatively using the CCK-8 assay. We performed an extraction-based cytotoxicity assay to evaluate the biocompatibility of the device under oral-like conditions. After ethylene oxide sterilization, the device was soaked in cell culture medium to prepare the extract solution. HPDLSCs and RAW264.7 macrophages were then incubated with the extract for 48 h, after which cell viability was measured using a CCK-8 assay. In vitro assessment of ultrasound-responsive release After the reservoirs of the device were filled with ZS, we sealed the surface with tridecanoic acid and soaked the drug chambers in an acidic solution at 37 °C. The application of US sequentially triggered the release of ZS from the first three drug chambers, whereas the release of ZS from the last three drug chambers was not triggered. For the release group, ultrasound was applied to actuate the first heating electrode (the ultrasound parameters were identical to those used for device triggering in other experiments), followed by the sequential actuation of the second and third electrodes to mimic repeated “on-demand” dosing from independent reservoirs. After each actuation cycle, the surrounding release medium was collected and then centrifuged, and the supernatant was analyzed with a UV200 spectrophotometer at 273 nm. The SenA concentration was calculated using the corresponding SenA standard curve. The nonrelease group was processed in parallel at matched sampling points to assess any leakage through the tridecanoic acid seal at 37 °C. In vivo experiments All animal experiments performed in this study adhered to the rules of the Animal Welfare Management Department, and approval was obtained from the Ethics Committee of the Affiliated Stomatological Hospital of Chongqing Medical University (License No. 2025 Ethics (010)). Thirty male New Zealand White rabbits, with a body weight of 3000 g and an age of 5 months, were distributed randomly into the following groups: C (control), P (periodontitis), P + S (periodontitis + SenA), P + ZS (periodontitis + SenA@ZIF-8), P + D (periodontitis + patch), and P + D+ZS (periodontitis + device baseplate + SenA@ZIF-8). The P + D+ZS group was treated with a US intensity of 2.5 W/cm², a duty cycle of 100%, an exposure time of 4 min, a coupling medium of medical ultrasound gel, and a distance of 1 cm. A periodontitis model was constructed through orthodontic ligature wire ligation and Porphyromonas gingivalis injection (purchased from ATCC 33277). An orthodontic ligature wire was tied at the cervical edge of the mandibular front teeth and placed near but not pressing the gingiva. Each week, the staff checked the degree to which the ligature fell off and administered a PG injection; this process lasted 6–8 weeks until obvious deep periodontal pockets and red and swollen gingiva with bleeding symptoms were observed. Silicone rubber impressions were made to establish the occlusal relation status. Hard plaster was poured into impressions, and a pure baseplate was manufactured using baseplate polymer via the box-and-pour method. Baseplate parts were produced through powder-and-pour casting technology. PZT was used as the energy supply source. A dry, quantified amount of SenA@Zif-8 was placed in the groove-shaped drug storage part at the ulnar end. Tridecanoic acid was used to seal the surface. Photographs (Nikon, Japan), thermal images (infrared thermal imager, Fotric288-L255; Fotric, China), and radiographs (digital radiographic X-ray system, DFM606; Angell, China) were captured throughout the treatment process. Indicators, including body weight, periodontal pocket depth, the gingival index, and the plaque index, were recorded. After 6 weeks of treatment, New Zealand rabbits were sacrificed by carbon dioxide inhalation. Whole blood samples were collected to conduct hematological, plasma, and enzyme activity analyses. Gingival tissue samples were tested for enzyme activity, and mandibular bone samples were fixed for micro-CT (Quantum GX3; Revvity, USA) analysis, H&E staining, and Masson’s trichrome staining. Visceral samples were fixed for H&E staining. More detailed descriptions and In Vivo Preliminary Study are provided in the Supporting Information. Statistical analysis All the data are shown as the means ± standard deviations (SDs), with standardized information on the analysis provided in the figure legends. All the data in this study were obtained from three or more independent experiments (n ≥ 3). Two-sample t tests were applied for comparisons between two treatment groups, whereas one-way ANOVA was used for comparisons among several treatment groups. Two-way ANOVA was adopted for comparisons among multiple treatment groups together with different treatment durations. *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001 are regarded as statistically significant. All the data were processed with GraphPad Prism software. Results and discussion In vivo preliminary study and device performance testing Preliminary research has indicated that a single dose of SenA@ZIF-8 resulted in very small changes in the serum total antioxidant capacity and CAT and SOD activities (Fig. 2A–C). In contrast, repeated drug administration significantly restored antioxidant enzyme activity; therefore, sustained, long-term drug delivery is needed for effective inflammation control in individuals with periodontitis. These findings are consistent with current clinical practice, in which repeated local delivery of agents such as minocycline is needed for disease management but is still invasive and inconvenient for patients. Nevertheless, for the sustained management of severe periodontitis after initial therapy, repeated local delivery of agents such as minocycline hydrochloride is needed [32–35]. Existing minocycline preparations depend on special devices for delivery into periodontal pockets, and patients have difficulty performing self-administration without professional help. Moreover, repeated pocket injection often leads to uncomfortable feelings, including swelling and pain, which reduce patient compliance and limit long-term self-management. These shortcomings emphasize the need for a precise, minimally invasive, and patient-friendly drug delivery strategy that allows the convenient at-home treatment of periodontitis. Accordingly, this design is intended to allow standardized, externally triggered dosing (potentially patient-operated with a portable ultrasound probe), thereby reducing logistical constraints associated with repeated chairside medication while maintaining precise spatiotemporal delivery [36]. The wireless power component combines a lead zirconate titanate piezoelectric transducer (PZT) to convert ultrasonic mechanical energy into electrical output. As confirmed by X-ray diffraction (XRD), the 2 × 2 × 1 mm³ PZT element shows an even grain shape and a main tetragonal phase, which matches its strong piezoelectric ability. The impedance analysis revealed a 1 MHz resonance frequency, which can realize effective acousto–electric energy conversion to provide power for downstream circuits and the heating electrode (Fig. 2D, Figure S1 and Figure S2). We measured the electrical output of the PZT element upon external ultrasonic stimulation to show that the ultrasonic wireless power transfer scheme is feasible and accurate. A 1 MHz alternating current (AC) was applied to drive an external ultrasound probe, which emits ultrasound at the same frequency, in this experiment. After the ultrasound waves travelled a certain distance, the PZT element successfully transformed the received mechanical energy into an AC electrical signal with the same frequency; the peak-to-peak voltage and current were approximately 10 V and 9 mA, respectively (Fig. 2E, F). Because the ultrasound waves propagated and attenuated in the medium, the output waveform of the PZT had a time delay of approximately 5.5 µs compared with the excitation signal, and its amplitude was attenuated at the same time as the excitation (Fig. 2F, inset). In the intended application situation, the external ultrasonic probe and the implanted PZT element are separated by human tissue. Therefore, we performed a finite element analysis (FEA) to evaluate the ultrasonic energy transmission efficiency across this medium. The simulation results, which are shown in Fig. 2G, H, indicated that under the given excitation conditions, the ultrasound waves maintained good directivity and were attenuated slightly after passing through 3 cm of tissue. Thus, ultrasound energy can effectively reach and activate the PZT element, which is located deep inside the tissue. We first rectified the AC signal produced by the PZT into a direct current (DC) signal to drive subsequent circuits, as displayed in Fig. 2I. We examined how the tissue depth, ultrasound power, and ultrasound incidence angle influence the rectified output voltage to systematically assess the controllability of this wireless power system. First, the output voltage displayed a two-stage dependence on the propagation depth; namely, it remained relatively stable within the first 1 cm, after which it gradually decreased with increasing depth (see Fig. 2J). Second, the peak output voltage was positively correlated with the transmitted ultrasonic power, as indicated in Fig. 2K. Third, the energy transfer efficiency reached a maximum value when the incident ultrasound wave was perpendicular to the PZT electrode surface, which resulted in the highest output voltage, and the output was minimized when the incident direction is parallel to the electrode surface, as shown in Fig. 2L. These findings confirm that using wireless ultrasonic drive, we can effectively control the operational state of the system at different tissue depths. The storage capacitor was used to store the rectified DC output. We conducted charging operations on 470 µF and 1000 µF capacitors to characterize its charging performance. Figure 2M shows that 5.0 V from these capacitors arrived within 130 s and 205 s, respectively. Therefore, the energy saved in these capacitors is applied to supply electric power for the subsequent circuit and the heating electrode. The heating electrode is a serpentine structure with a 100 μm line width, and it is composed of a 40 nm chromium adhesion layer and a 200 nm copper layer (Fig. 2N). First, we evaluated its electrothermal performance. Therefore, under different electrical drive powers, the electrode temperature correspondingly increased and stabilized at approximately 10 s; both the heating response time and the cooling response time were less than 5 s (Fig. 2O). Infrared thermography verified that the temperature increase was precisely localized inside the electrode area (Fig. 2P). In terms of biocompatibility, polydimethylsiloxane (PDMS) was used to encapsulate the system surface. Hence, we compared the temperature changes during five heating‒cooling cycles with progressively increasing power between bare electrodes and PDMS-encapsulated electrodes to study the influence of encapsulation on thermal performance. The results showed that PDMS encapsulation had a negligible effect on the final steady-state temperature (Fig. 2Q, R) and led to a more stable surface temperature distribution (Fig. 2S). However, because the device contained a PDMS layer, an extra heat capacity was introduced, which caused a small increase in the thermal response time (Fig. 2T). Design strategies and characterization of SenA@ZIF-8 Senkyunolide A (SenA) has very strong anti-inflammatory and antioxidant activities [37–39], and ZIF-8 acts as a high-efficiency carrier for drug encapsulation and controlled release [40, 41]. SenA@ZIF-8 was produced by adsorbing SenA onto premanufactured ZIF-8 nanoparticles [41] (Fig. 3A). We constructed a UV–Vis calibration curve for SenA at 273 nm to quantify the incorporation of SenA into ZIF-8, and obtained a linear regression equation of y = 0.0116x + 0.0545 (R² = 0.9995), indicating reliable quantification within the tested range (Figure S3). The content of free (unloaded) SenA in the postloading supernatant was measured and used for mass-balance calculations of the encapsulated fraction. Based on the results from three independent replicates (n = 3), the mass of SenA loaded into ZIF-8 was 2.498 ± 0.041 mg (with 10 mg of ZIF-8 used as the carrier input), corresponding to a loading capacity (LC) of 19.989 ± 0.262 wt% and an encapsulation efficiency (EE) of 49.968 ± 0.816% (Table S3). The low replicate variability (CV ≈ 1–2%) indicates good reproducibility of the loading process. From a materials perspective, these results are consistent with the use of ZIF-8 as a drug carrier: ZIF-8 is widely reported to have high porosity and a large surface area with good stability, which supports the adsorption/encapsulation of small-molecule therapeutics and has made it a widely studied MOF platform in drug delivery. Functionally, achieving ~20 wt% LC ensures that therapeutically relevant payloads can be delivered with a practical carrier mass, which is important for confined local administration sites such as periodontal pockets. Moreover, an EE of ~50% suggests that approximately half of the input SenA is effectively incorporated under the current loading conditions; this level is sufficient for our patch-based dosing design but also indicates room for further optimization (e.g., solvent composition, loading concentration, or incubation time) if needed for future scale-up studies. Scanning electron microscopy (SEM) revealed that pure ZIF-8 has a uniform dodecahedral shape (Fig. 3B), and dynamic light scattering (DLS) revealed that its average hydrodynamic diameter was 154.35 nm (Fig. 3C). After drug loading, SenA@ZIF-8 presented a slightly increased particle size and a more rounded outer form. The Fourier transform infrared spectroscopy (FTIR) spectra retained the characteristic peaks of ZIF-8, indicating that the completeness of the framework was maintained after SenA was added (Fig. 3D). A representative photograph of the as-prepared SenA@ZIF-8 (Fig. 3E) further confirms the successful formation of a uniformly dispersed product at the macroscopic level, complementing the microscopic characterization results. Material characterization and in vitro antioxidant and osteogenic activities of SenA@ZIF-8. (A) Synthesis of SenA@ZIF-8. (B) SEM images of ZIF-8 (left panel) and SenA@ZIF-8 (right panel). (C) DLS analysis of ZIF-8. (D) FTIR spectra. (E) Photograph of the as-prepared ZS dispersion. (F, G) Cytotoxicity at 48 h and proliferation at 1/3/5 days in hPDLSCs treated with ZIF-8, SenA, or SenA@ZIF-8. (H–J) Antioxidant actions opposing H₂O₂-triggered oxidative stress for 3 h, including the antioxidant-mediated restoration of the mitochondrial membrane potential and ROS levels. (K, L) Images of ALP staining at 7 days and Alizarin Red staining at 21 days. (M–P) Quantitative analysis of the results shown in I–L. (Q–T) mRNA expression of oxidative stress-related genes. The data are presented as the means ± SDs; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 Toxicity, antioxidant effects and osteogenic differentiation-inducing effects of SenA@ZIF-8 on hPDLSCs First, we evaluated the cytotoxicity of SenA, ZIF-8, and SenA@ZIF-8 toward human periodontal ligament-derived stem cells (hPDLSCs) by performing a CCK-8 assay. SenA exhibited negligible cytotoxicity at concentrations up to 100 µg/mL. In comparison, ZIF-8 and SenA@ZIF-8 were well tolerated at concentrations below 60 µg/mL, whereas higher concentrations resulted in a marked reduction in cell viability. Therefore, concentrations of 40 and 60 µg/mL were chosen for the subsequent experiments. At both of these concentrations, all three preparations promoted hPDLSC proliferation in a dose-dependent manner, and no short-term cytotoxicity was detected (Fig. 3F and G). Periodontitis is a complicated, nonlinear chronic inflammatory disease that leads to the generation of excessive amounts of and unbalanced amounts of ROS in hPDLSCs. High ROS levels directly damage cells, and the inflammatory environment caused by ROS represses the expression of osteogenic markers, thus impairing their calcification and osteogenic functions [42, 43]. hPDLSCs were challenged with H₂O₂ to construct a model of oxidative stress related to periodontitis [44]. An IC₅₀ of ~0.25 mM was measured (Figure S4 and Figure S5) and was applied in subsequent stimulation experiments. Therefore, compared with ZIF-8, SenA@ZIF-8 more effectively restored the viability of H₂O₂-stimulated hPDLSCs, with the greatest protective effect observed at 60 µg/mL (Fig. 3H). Therefore, 60 µg/mL was selected as the concentration for subsequent studies involving hPDLSCs. An analysis of the mitochondrial membrane potential revealed that H₂O₂ resulted in a notable loss of red fluorescence; this finding indicated mitochondrial depolarization and early apoptosis (Fig. 3I, M). Pretreatment with SenA@ZIF-8 significantly restored mitochondrial fluorescence, and this effect was greater than that of ZIF-8 or SenA alone. Consistent with these findings, after H₂O₂ exposure, intracellular ROS levels increased substantially, whereas SenA@ZIF-8 pretreatment markedly suppressed ROS accumulation (Fig. 3J, N). Reverse transcription‒quantitative real-time PCR (RT‒qPCR) further revealed that oxidative stress downregulated the expression of antioxidant genes, including CAT, GPX1, SOD-1, and HO-1, whereas compared with ZIF-8 or SenA alone, SenA@ZIF-8 significantly upregulated their expression (Fig. 3Q–T). Therefore, these results prove that SenA@ZIF-8 strengthens cellular antioxidant defenses and provides better protection against ROS-induced injury. One core function of periodontal ligament stem cells is their ability to undergo calcification and osteogenesis [45–47]. Persistent chronic inflammation disrupts the osteoclast–osteoblast coupling relationship, ultimately causing alveolar bone destruction and periodontal pocket formation [48]. Medical biomaterials are needed to recover the osteogenic function of periodontal ligament stem cells [49, 50]. We evaluated osteogenic differentiation at different stages by measuring Alkaline phosphatase (ALP) activity at the early stage of osteogenic induction and performing alizarin red S (ARS) staining at the late stage to evaluate mineralized matrix formation. As shown in Fig. 3K and O, ZIF-8 partially restored ALP activity in cells under oxidative stress, whereas SenA had a more pronounced effect. However, SenA@ZIF-8 did not show a significant advantage over the control treatment at this early stage. In contrast, at the later stage of differentiation, ARS staining revealed a markedly different trend. H₂O₂-treated cells showed minimal mineralized nodule formation, whereas SenA@ZIF-8 significantly increased mineralization, even exceeding that in the untreated control group (Fig. 3L, P). This apparent discrepancy between early ALP activity and late-stage mineralization suggests a time-dependent osteogenic regulatory mechanism. Specifically, oxidative stress initially suppresses early osteogenic activity, and the antioxidant effect of SenA@ZIF-8 primarily involves reducing intracellular ROS levels and restoring cellular homeostasis rather than immediately increasing the expression of early differentiation markers. As the intracellular microenvironment improves over time, the sustained synergistic release of SenA and Zn²⁺ ions from ZIF-8 may promote osteogenic maturation and extracellular matrix mineralization. Therefore, SenA@ZIF-8 has a delayed but more pronounced pro-osteogenic effect characterized by early-stage cytoprotective and redox-regulating functions, followed by increased late-stage mineral deposition. This two-stage regulatory behavior highlights its ability to treat oxidative stress-associated bone defects. However, the specific regulatory mechanism needs further verification. Anti-inflammatory function of SenA@ZIF-8 in RAW264.7 cells Periodontal bone loss caused by periodontitis is a type of inflammatory bone loss [51]. Therefore, controlling inflammation is another key target during periodontitis treatment [42, 52, 53]. RAW264.7 macrophages were used to assess the anti-inflammatory activity of SenA@ZIF-8 (Fig. 4A). First, cytocompatibility was measured using a CCK-8 assay. When the SenA@ZIF-8 concentration reached 40 µg/mL, cell viability remained above 80% (Figure S6); hence, 30 µg/mL was chosen as the concentration for subsequent experiments with RAW264.7 cells. An analysis of intracellular uptake showed that the intracellular SenA content was significantly higher in the SenA@ZIF-8 group than in the free SenA group (Fig. 4B). In fact, ZIF-8-mediated delivery resulted in an approximately twofold increase in the intracellular SenA level. These results indicate that SenA@ZIF-8 greatly increases the cellular uptake of SenA by RAW264.7 cells. However, SenA@ZIF-8 is poorly soluble in aqueous buffer and has a strong affinity for lipid membranes. We performed multiple PBS washes to remove loosely bound SenA prior to lysis. However, some SenA@ZIF-8 may remain attached to the cell surface. Therefore, the quantified “intracellular” SenA likely includes membrane-bound fractions. Despite this limitation, the functional benefits observed with SenA@ZIF-8 strongly suggest effective cellular uptake and bioactivity. Next, the anti-inflammatory effects were evaluated in RAW264.7 cells stimulated with lipopolysaccharide (LPS). Immunofluorescence staining confirmed that LPS induced a typical M1 phenotype in RAW264.7 cells, with significantly increased CD86 expression and low CD206 levels. Treatment with SenA and ZS markedly reversed this change, reducing CD86 expression and increasing CD206 expression. Notably, the ZS group exhibited the strongest shift toward the M2 phenotype, while the group treated with ZIF-8 alone showed only a modest effect. The results of the quantitative analysis confirmed a significant increase in CD86 expression and no increase in CD206 expression compared with those in the control group (Fig. 4C–E). RT‒qPCR analysis revealed that SenA could downregulate the mRNA expression of the M1 macrophage-related cytokines IL-1β, IL-6, iNOS, and TNF-α, whereas SenA@ZIF-8 further inhibited the expression of these proinflammatory markers (Fig. 4F–I). Moreover, compared with the ZIF-8 or SenA groups, the SenA@ZIF-8 group showed significantly upregulated expression of M2 macrophage-related genes (Arg-1, IL-10, and TGF-β) (Fig. 4J–L), indicating a transformation to an anti-inflammatory macrophage phenotype. Consistently, LPS stimulation clearly increased extracellular nitric oxide (NO) production, whereas ZIF-8, SenA, and SenA@ZIF-8 reduced NO release, with the SenA@ZIF-8 group showing the lowest NO level (Fig. 4M). ELISA further confirmed that SenA and SenA@ZIF-8 could decrease the LPS-induced secretion of IL-6 and TNF-α, and the inhibitory effect of SenA@ZIF-8 was the strongest (Fig. 4N, O). Therefore, these results prove that SenA@ZIF-8 can increase intracellular drug delivery and can exert powerful immunomodulatory effects by suppressing M1 macrophage polarization and promoting M2 polarization, thus effectively reducing inflammatory responses related to periodontitis. In vitro anti-inflammatory activity of SenA@ZIF-8. (A) Schematic of the SenA@ZIF-8-mediated regulation of LPS-induced RAW264.7 macrophage polarization. (B) Intracellular SenA levels detected following treatment with free SenA or ZIF-8-mediated delivery of SenA. (C–E) Immunofluorescence staining of RAW264.7 cells for markers of M1/M2 macrophages and the results of the quantitative analysis performed 24 h after LPS stimulation. (F–L) Expression of inflammatory and polarization-related markers detected 6 h after LPS stimulation. (M–O) Suppression of LPS-induced NO, IL-6, and TNF-α release at 24 h after LPS stimulation. The data are presented as the means ± SDs; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 Transcriptome sequencing analysis of RAW264.7 cells RAW264.7 macrophages were cocultured with ZIF-8, SenA, or SenA@ZIF-8 and subsequently stimulated with LPS for transcriptome sequencing (Fig. 5A). The expression of eleven genes (Il10, Upp1, Lcn2, Il36a, Procr, Egr2, Akap12, Lad1, Mmp13, Lipg, and Fabp4) showed consistent changes across all treatment groups; therefore, this finding indicates a shared immunomodulatory signature. We compared the SenA@ZIF-8 group and the free SenA group (LPS + ZS vs. LPS + SenA) to determine the specific effects of ZIF-8-mediated delivery. A total of 7,264 genes were upregulated, and 1,174 genes were downregulated (Fig. 5B). Gene Ontology (GO) and KEGG pathway analyses revealed significant enrichment in immune- and inflammation-related processes, including “inflammatory response,” “immune receptor activity,” and “regulation of immune response” (Fig. 5C, D). Notably, the results of the KEGG analysis revealed that the TNF, NF-κB, and PI3K–Akt signaling pathways were dominant nodes, suggesting that SenA@ZIF-8 exerts its immunoregulatory effects primarily through these core inflammatory cascades. Among the commonly regulated genes, several have been directly implicated in periodontitis pathogenesis. Interleukin-10 (IL-10), which is a key immunosuppressive cytokine that limits excessive inflammation and mitigates host tissue damage [54], was upregulated, consistent with an anti-inflammatory macrophage phenotype [55]. Lipocalin 2 (LCN2) is a macrophage-derived glycoprotein involved in immune regulation and iron homeostasis; it has been reported to impair osteogenic differentiation when its expression is aberrantly elevated during periodontitis [56, 57]. Matrix metalloproteinase-13 (MMP-13) is a collagenase associated with extracellular matrix degradation and osteoclast-mediated bone resorption, and it is a well-established contributor to periodontal tissue destruction [58, 59]. Fatty acid binding protein 4 (FABP4) is a macrophage-expressed lipid-binding protein that functions as a proinflammatory mediator and has recently been linked to the systemic vascular complications of periodontitis [60, 61]. Different approaches to the transcriptome sequencing analysis of RAW264.7 cells. (A) Venn diagram of the treatment groups. (B) Volcano plot of differentially expressed genes in RAW264.7 cells following LPS stimulation and treatment with ZS or SenA. (C) KEGG enrichment analysis of RAW264.7 cells treated with ZS and SenA following LPS stimulation. (D) GO enrichment analysis of RAW264.7 cells treated with ZS and SenA following LPS stimulation Importantly, our transcriptomic data provide an additional mechanistic bridge through an osteoimmunomodulatory axis. The RNA-seq analysis revealed the significant enrichment of pathways and terms related to inflammatory regulation, including TNF signaling, NF-κB signaling, cytokine–cytokine receptor interaction, and immune response GO categories, as well as PI3K–Akt/MAPK nodes. These pathways are core determinants of macrophage inflammatory output and influence the osteogenic performance of mesenchymal/stromal lineages under inflammatory conditions. Since proinflammatory cytokines (e.g., TNF-α/IL-6/IL-1β) can inhibit the osteogenic differentiation of periodontal ligament stem cells, the suppression of these inflammatory cascades is expected to indirectly favor osteogenesis by attenuating inflammation. In parallel, macrophage polarization strongly regulates bone regeneration; proresolving, IL-10-associated macrophage responses have been reported to increase MSC osteogenesis and bone formation in an IL-10-dependent manner. Therefore, integrating the transcriptomic pathway signatures with our ALP/ARS phenotypes supports a model in which SenA@ZIF-8 promotes osteogenesis not only through direct redox and Zn²⁺-linked osteoinduction in hPDLSCs but also by reshaping inflammatory signaling programs that otherwise suppress mineralization during chronic periodontal inflammation. Collectively, the coordinated regulation of inflammation- and bone-remodeling-related genes, together with the enrichment of TNF, NF-κB, and PI3K–Akt pathways, suggests that SenA@ZIF-8 may modulate macrophage immune responses through multiple signaling networks. These transcriptomic findings are consistent with the observed suppression of M1 polarization, enhancement of M2 polarization, and reduction of inflammatory mediator production at the cellular level. However, it should be noted that the current transcriptomic data provide pathway-level evidence only, and do not directly verify the activation or inhibition status of key signaling proteins. In particular, protein-level validation of representative pathway molecules, such as p- NF-κB and p-IκBα, was not performed in this study. Therefore, the proposed mechanism should be interpreted as preliminary. Future studies will incorporate Western blotting, immunofluorescence, and targeted gene validation to more precisely define the molecular pathways underlying SenA@ZIF-8–mediated regulation of inflammation, oxidative stress, and osteogenesis. Preparation, function and biocompatibility of device bases for in vivo therapy Heating electrodes were combined into custom-made circuits and placed together into a wearable bottom plate to facilitate therapy in living subjects (Fig. 6A). Silicone molds of the front teeth of New Zealand rabbits were generated to record bite relationships, and resin bottom plates were made from plaster molds (Fig. 6B). The control circuit was designed to fit the shape of the bottom plate (Fig. 6C), printed, and inserted into the bottom plate through a powder-casting method (Fig. 6D). Therefore, the control circuit was placed on the tongue-side of the bottom plate, while six drug storage containers were equally arranged on the lip-side, middle-to-far, and tongue-side surfaces of the two front teeth. Thus, different heating electrodes were started under input voltages from 5.0 to 4.2 V (constant current: 20 mA) to allow controllable thermal action. Hence, the PZT power component was connected to provide wireless electrical energy. SenA@ZIF-8 nanoparticles were centrifuged, dried, placed in the drug storage containers, and sealed with a thin tridecanoic acid layer to prevent premature release. Experimental workflow and structural characterization of the device baseplate in New Zealand White rabbits. (A) Flowchart of device baseplate fabrication. (B) Physical images of the pure baseplate. (C) Schematic and physical images of the device circuit board. (D) Physical views of the device baseplate. (E–H) Working drawings illustrating rabbit incisors. (I–L) X-ray images of rabbit heads; the white arrow represents the heating electrode. (M–N) In vivo thermal imaging when the device was working Photographs that show anterior teeth with patches and without patches revealed excellent retention and minimal gingival irritation (Fig. 6E, F). Although some incisors were covered, the baseplate could fit smoothly and remain stable (Fig. 6G, H). Radiographic imaging further confirmed that the device was placed securely (Fig. 6I–L). Because the baseplate fitted tightly, the entire circuit was enclosed by gingival tissue in the cervical region (Fig. 6J). A modified baseplate with a mesially deflected PZT module was produced to elucidate the spatial relationships among the heating electrode, drug reservoir, and teeth. X-ray imaging revealed that the labial heating electrode was closely adhered to the cervical region, which corresponded to the periodontal pocket location; therefore, these findings confirm that both the drug reservoir and the heating electrode are positioned accurately for targeted delivery (Fig. 6L). We used infrared thermography to monitor the in vivo thermal performance of the device. The baseline surface temperature of the rabbits without this device was 31.8 °C (Figure S7). Therefore, after we placed the baseplate, the maximum surface temperature slightly decreased to 29.8 °C (Fig. 6M). Upon ultrasound activation, the local temperature increased quickly to 43.2 °C (Fig. 6N), which was higher than the melting point of tridecanoic acid, and these conditions resulted in the on-demand release of SenA@ZIF-8 from the drug reservoirs. Hence, these experimental results confirm that the device can accomplish precise, localized heating, which is sufficient for controlled drug release, and that this heating does not result in excessive thermal exposure to the rabbits. In vivo periodontal tissue regeneration Together, the pilot dosing study and the clinical feasibility considerations provide a clear translational rationale for device-based therapy: manual repeated dosing may be effective but is impractical, whereas our patch aims to deliver the functional equivalent of repeated dosing through single placement + on-demand activation. We evaluated the therapeutic effect of the intelligent SenA@ZIF-8 delivery system in a rabbit periodontitis model induced by orthodontic wire ligation plus P. gingivalis injection (Fig. 7A). Changes in the clinical indicators of New Zealand White rabbits treated with the device baseplate. (A) In vivo experimental workflow. (B, C) Toxicity of tridecanoic acid in hPDLSCs/Raw264.7 cells with/without US stimulation measured at 48 h. (D, E) Cell viability after 48 h of incubation with medium extracts from the PDMS-encapsulated device. (F) In vitro ultrasound-responsive ZS release from the tridecanoic acid patch. (G–J) Longitudinal changes in body weight, periodontal pocket depth, gingival index, and plaque index during modeling and 2, 4, and 6 weeks after treatment. (K) Representative photographs of periodontal pocket depth before and 6 weeks after treatment. The data are presented as the means ± SDs; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 Cytocompatibility was evaluated in both hPDLSCs and RAW264.7 cells with and without ultrasound exposure to assess the biosafety of tridecanoic acid under trigger-relevant conditions. In both cell types, ultrasound treatment alone did not reduce cell viability, confirming the safety of the applied acoustic parameters within the experimental window. Across a broad concentration range (10–640 µg/mL), tridecanoic acid exhibited minimal cytotoxicity, with comparable cell viability observed between − US and + US conditions. At the highest concentration tested (1280 µg/mL), a moderate decrease in viability was observed in hPDLSCs upon ultrasound stimulation, whereas RAW264.7 cells remained largely unaffected (Fig. 7B, C). This difference may reflect the greater sensitivity of stem cells than macrophages to environmental stress . Importantly, even under these conditions, cell viability remained within an acceptable range, indicating no overt cytotoxicity. From a translational perspective, we must emphasize that the tridecanoic acid in our system functions as a confined phase-change barrier rather than a freely administered agent. Therefore, the in vitro evaluation represents a worst-case exposure scenario. In vivo, cellular exposure is expected to be significantly lower because of spatial confinement within the reservoir, the transient and localized nature of the melting process, and dilution by the periodontal microenvironment. PDMS coatings significantly limit the moisture-induced degradation of electronic implants [62]. In addition, the results of cytotoxicity assays showed that the viability in both hPDLSCs and RAW264.7 cells remained above ~95% after 48 h of exposure to the device extracts (Fig. 7D, E), indicating negligible leachable toxicity. These results demonstrate that the PDMS-encapsulated device does not release harmful byproducts under simulated oral conditions. Overall, the PDMS packaging of the device provides effective waterproofing and electrical insulation, ensuring safe and durable operation in the oral environment. We quantified SenA release from the tridecanoic acid-sealed patch in an acid solution at 37 °C upon ultrasound actuation compared with a matched no-ultrasound control to experimentally validate the ultrasound-triggered mechanism. As shown in Fig. 7F, the No-US control remained at baseline across three sequential sampling points, indicating negligible leakage of SenA@ZIF-8 at 37 °C and confirming that the tridecanoic acid layer provides effective sealing when incubated at physiological temperature. In contrast, the US-actuated group exhibited a distinct stepwise increase in measurable SenA levels after each sequential actuation (1–3 times), reaching ~1.1 µg/mL after the first actuation and increasing to ~3.5 µg/mL after the third actuation. This progressive increase is consistent with on-demand release from independently addressable reservoirs, where ultrasonic energy enables the heating electrode to locally melt the phase-change barrier, thereby allowing the internal load to be released. Importantly, the absence of a detectable SenA signal in no‑ultrasound control under the same medium and sampling workflow supports that the observed release is triggered and controlled rather than driven by passive diffusion, thereby providing direct experimental validation of ultrasound responsiveness. Here, tridecanoic acid serves solely as a heat-sensitive gate. With a melting point of ~41–42 °C, tridecanoic acid is solid at physiological temperature and prevents any premature SenA release. Only after ultrasound heating does the tridecanoic acid melt, releasing SenA@ZIF-8 into the pocket. Because the pH of the gingival fluid is slightly alkaline [63–66], the ZIF-8 framework remains essentially intact. Actual SenA release occurs intracellularly, as endocytosed ZIF-8 encounters acidic lysosomes (pH ~4.5–5.0), where it rapidly degrades and liberates its payload. Future studies will incorporate the direct quantification of the residual payload to further refine the understanding of release efficiency and dosing precision. We divided the rabbits into six groups: control (C), periodontitis (P), periodontitis + SenA (P + S), periodontitis + SenA@ZIF-8 (P + ZS), periodontitis + patch (P + D), and periodontitis + device-based SenA@ZIF-8 delivery (P + D + ZS) to evaluate the function of the device in vivo. We monitored body weight every week (Fig. 7G) and measured the periodontal pocket depth, gingival index, and plaque index once every two weeks (Fig. 7H–J). In addition, compared with those in the periodontitis (P) and periodontitis + patch (P + D) groups, the weight of the animals that received the device containing SenA@ZIF-8 (P + D + ZS) significantly increased, suggesting that this beneficial whole-body effect is linked to effective control of local inflammation. Next, we evaluated local periodontal recovery. Manual administration of SenA or SenA@ZIF-8 (P + S or P + ZS) decreased the periodontal pocket depth, whereas device-mediated delivery of SenA@ZIF-8 (P + D + ZS) resulted in the greatest improvement, with the posttreatment pocket depth approaching the control level (Fig. 7K and Table S4). The gingival index decreased slightly in the periodontitis + SenA (P + S) group and decreased to grade 1 (mild inflammation) in both the periodontitis + SenA@ZIF-8 and periodontitis + device-based SenA@ZIF-8 delivery (P + ZS and P + D + ZS) groups, as evidenced by a slight color change, mild edema, and no bleeding during probing (Figure S8 and Table S5). Plaque accumulation decreased sequentially from the periodontitis + SenA (P + S) group to the periodontitis + SenA@ZIF-8 (P + ZS) group and reached near-control levels in the periodontitis + device-based SenA@ZIF-8 (P + D + ZS) group, with no statistically significant difference from those in healthy animals (Figure S9 and Table S6). Hence, these results prove that device-based SenA@ZIF-8 delivery (P + D + ZS) can achieve long-lasting, local, and minimally invasive treatment, leading to better control of periodontal inflammation and tissue regeneration than manual drug administration. This device provides a practical and patient-friendly method for the long-term management of periodontitis. In vivo bone regeneration and systemic anti-inflammatory effects After 6 weeks of treatment, we euthanized New Zealand rabbits and collected their mandibles, gingival tissues and whole blood for biological assessments. The micro-CT analysis of anterior mandibular teeth revealed that wearing a patch alone (P + D) did not cause severe alveolar bone resorption. In contrast, compared with free SenA (P + S), SenA@ZIF-8 (P + ZS) clearly promoted alveolar bone repair; compared with free SenA (P + S), SenA@ZIF-8 (P + ZS) had better osteogenic ability, consistent accordance with the ALP and ARS results obtained in vitro (Fig. 8A, B). Notably, compared with manual application of SenA@ZIF-8 (P + ZS), the device-mediated delivery of SenA@ZIF-8 (P + D + ZS) further increased the degree of bone regeneration, as indicated by a greater bone volume fraction (BV/TV) and decreased trabecular separation (Tb. Sp) (Fig. 8C–F). The trabecular number (Tb. N) was the only parameter that significantly improved compared with that in the periodontitis (P) group, thus highlighting that the reconstruction of the microarchitecture is sensitive to continuous, local therapy. These results indicate that minimally invasive, device-based delivery provides a more suitable microenvironment for bone regeneration after periodontitis than manual injection. The histological analysis further confirmed these outcomes. H&E and Masson’s trichrome staining of decalcified alveolar bone sections showed that inflammatory cell infiltration gradually decreased and that collagen fiber deposition increased in the SenA, SenA@ZIF-8 and device-mediated delivery of SenA@ZIF-8 (P + S, P + ZS, and P + D + ZS) groups (Fig. 8G, H). Among all these groups, device-delivered SenA@ZIF-8 (P + D + ZS) had the most significant anti-inflammatory and pro-regenerative effects, and SenA loaded onto ZIF-8 (P + ZS) continued to perform better than free SenA (P + S). Combined therapeutic efficacy of ZS and the baseplate device in New Zealand White rabbits with periodontitis after treatment. (A) Micro-CT three-dimensional reconstruction and sagittal sections of mandibular incisors after different treatments, with the red line indicating the distance between the alveolar bone crest (ABC) and the cementoenamel junction (CEJ). (B) Quantification of the ABC–CEJ distance. (C–F) Quantitative analysis of alveolar bone morphometric parameters. (G–H) Representative images of Masson’s trichrome and H&E staining of periodontal tissues. (I–L) Posttreatment peripheral blood leukocyte profiles, including total white blood cell, neutrophil, lymphocyte, and eosinophil counts. (M–R) Antioxidant activity, including the total antioxidant capacity, catalase activity, and superoxide dismutase activity, in serum and gingival tissues. The data are presented as the means ± SDs; *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 Because periodontitis causes inflammation both locally and systemically, we also measured the levels of inflammatory indicators in peripheral blood. Patients with severe periodontitis have higher levels of proinflammatory mediators in the blood and greater numbers of neutrophils. Conversely, successful localized periodontal therapy can reduce the levels of systemic inflammatory markers [51, 67–69]. The complete blood count showed that total leukocyte and neutrophil counts were markedly lower in all treatment groups than in the periodontitis (P) group, and the group in which the device was used to deliver SenA@ZIF-8 (P + D + ZS) had values closest to those of the healthy control (C) group (Fig. 8I, J). Moreover, lymphocyte and eosinophil counts decreased significantly in the SenA@ZIF-8 (P + D + ZS) group (Fig. 8K, L); therefore, device-mediated SenA@ZIF-8 delivery (P + D + ZS) not only inhibited local periodontal inflammation but also reduced the systemic inflammatory load. Furthermore, no systemic toxicity was detected with any of the treatment methods (Figures S10–16). Periodontitis is also associated with damaged antioxidant defenses both locally and systemically [70–72]. After treatment, the total antioxidant capacity and CAT and SOD activities were clearly increased in the gingival tissue and serum. Manual administration of SenA@ZIF-8 (P + ZS) produced better outcomes than free SenA (P + S); hence, among all the groups, the greatest recovery of the antioxidant capacity was obtained with device-mediated SenA@ZIF-8 delivery (P + D + ZS) (Fig. 8M–R). Collectively, these results prove that compared with manual drug administration, device-delivered SenA@ZIF-8 (P + D + ZS) exerts better osteogenic, anti-inflammatory, and antioxidant treatment effects. Therefore, this sustained, minimally invasive method can effectively advance periodontal tissue regeneration and simultaneously reduce systemic inflammatory stress, thus highlighting its strong translational potential for long-term periodontitis management. The superior performance of this device-assisted therapy (P + D + ZS) compared with traditional manual administration methods (P + ZS) can be attributed to significant improvements in dose control and reduced tissue trauma. By utilizing focused ultrasound technology, the device can precisely localize drug release to the periodontal lesion site and trigger release on demand. Thus, each ultrasound pulse instantly releases a precise amount of ZS at the exact location and time when it is needed, thereby establishing a stable drug concentration in the local area. In contrast, traditional manual gel or paste administration methods interfere with periodontal ligament regeneration. These combined effects ensure that an effective therapeutic concentration of the drug is maintained in the local area while minimizing the trauma associated with administration. This treatment ultimately manifests as clinically observed reductions in inflammatory marker levels and improvements in bone regeneration [73]. Interanimal variability in drug release was not quantified in this study; however, the consistent therapeutic outcomes across animals suggest reproducible and effective dosing. Future studies will include quantitative assessments of release variability to further improve dosing precision. Periodontitis management is currently guided by staging/grading and stepwise therapeutic frameworks, where initial nonsurgical periodontal therapy is followed by re-evaluation and site-specific management of residual pockets. In this context, our device is positioned for patients with Stage II–III periodontitis (typically moderate–severe disease) who, after initial therapy, present with isolated residual pockets requiring further localized intervention and close maintenance [74]. Local adjunct therapy is clinically used to deliver high drug concentrations directly into periodontal pockets while reducing unnecessary systemic exposure. For example, controlled-release local antimicrobials (e.g., minocycline formulations) have been reported to improve clinical outcomes when used as adjuncts to subgingival instrumentation, and professional guidelines endorse locally delivered sustained/controlled-release antimicrobials as an adjunct therapy in periodontitis management [75–77]. Our patch extends this concept by improving precision and retention within the pocket and by enabling repeatable on-demand dosing from a single placement (proposed default: ultrasound triggers every 48–72 h during the active phase), thereby reducing the need for repeated chairside intrapocket administrations, which is particularly relevant for patients with poor compliance, limited access to care, or where systemic antibiotics are undesirable. Despite the promising therapeutic efficacy observed in this study, several limitations should be acknowledged. First, the in vivo evaluation was conducted over a relatively short period, mainly reflecting early-stage anti-inflammatory and osteogenic responses, whereas the long-term stability of periodontal regeneration requires further investigation. Although the materials used in this system, including ZIF-8 and tridecanoic acid, exhibit favorable biocompatibility, the long-term degradation behavior and clearance of the integrated device in the complex oral environment remain to be fully elucidated. Second, the ultrasound-triggered drug delivery strategy, which is effective under controlled experimental conditions, may face challenges in clinical application because of the variability in ultrasound delivery and user operation. From a translational perspective, future efforts will focus on developing biodegradable device systems to enable complete resorption after therapy, as well as optimizing the wireless power and control system through the miniaturization of piezoelectric components and integration with portable ultrasound devices [78–80]. These improvements are expected to increase patient convenience, treatment consistency, and clinical feasibility. Another limitation of this study is the lack of a direct microbiological evaluation. Periodontitis is fundamentally a biofilm-driven disease, and parameters such as the P. gingivalis burden, plaque biofilm mass, and microbiome composition were not assessed in the present study. Future studies will incorporate microbiological endpoints, including the quantification of the bacterial load and biofilm analyses, to comprehensively evaluate the interaction between antimicrobial effects and the modulation of the host response. Conclusions Controlling inflammation through repeated drug administration after initial therapy remains a persistent clinical challenge in patients with severe periodontitis. Leveraging the high adsorption capacity of ZIF-8 for small-molecule drugs, we engineered SenA@ZIF-8 nanoparticles and integrated them into a wearable delivery device that enables minimally invasive, convenient, and precise periodontal therapy. This system directly addresses poor patient compliance associated with long-term, clinic-based follow-up injections. SenA intrinsically exhibits anti-inflammatory and antioxidant activities. When loaded into ZIF-8, it achieves higher intracellular bioavailability and synergizes with the framework to enhance osteogenic outcomes through combined immunomodulatory and redox-regulating effects. Device-mediated delivery further enables sustained, localized dosing without disrupting newly formed periodontal ligament attachments, thereby creating a favorable microenvironment for periodontal regeneration. Mechanistically, this platform achieves the coordinated suppression of inflammation, attenuation of oxidative stress, and promotion of M2 macrophage polarization, collectively facilitating alveolar bone regeneration and the ordered reconstruction of periodontal ligament fibers. In addition to therapeutic efficacy, the wearable design enables precise at-home administration, minimizing procedural trauma and reducing dependence on repeated clinical interventions. In the future, further miniaturization of the device will enable truly microdevice-based periodontal therapy, increase patient autonomy and reduce the clinical workload. This strategy holds strong translational potential as an adjunct delivery platform following foundational treatment for moderate-to-severe periodontitis. Supplementary Information Supplementary Material 1. Acknowledgements The writers express thanks to every study participant and every collaborator for carrying out data collection work. Hence, the authors are full of gratitude to School of Physics, University of Electronic Science and Technology of China, Chengdu, 611731, China for the support which is provided to them. Author contributions B.P: Conceptualization, Methodology, Investigation, Formal analysis, Writing-Original Draft. J.Y: Investigation, Formal analysis. R.L: Investigation, Formal analysis, Writing-Original Draft. XY.X: Resources, Funding acquisition. HW.S: Conceptualization, Writing - Review & Editing, Funding acquisition. JL.S: Resources, Funding acquisition. XM.Z: Resources, Supervision, Writing - Review & Editing, Funding acquisition. All authors reviewed the manuscript. Funding This research task was financially supported by the National Natural Science Foundation of China (82470988), the National Natural Science Foundation of China (Grant No. U22A20314), the China Postdoctoral Science Foundation Project (2024M763912), Outstanding Youth Fund of Chongqing Natural Science Foundation (CSTB2023NSCQ-JQX0006), the Chongqing Natural Science Foundation General Project (CSTB2024NSCQ-MSX0445),and Key Project of Science and Technology Research Plan of Chongqing Municipal Education Commission (KJZD-K202500412). Data availability No datasets were generated or analysed during the current study. Declarations 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. Bo Peng, Jing Yang and Rui Lin contributed equally to this study. Contributor Information Xinyu Xue, Email: xuexinyu@uestc.edu.cn. Huawei Shen, Email: huaweishen2022@163.com. Jinlin Song, Email: songjinlin@hospital.cqmu.edu.cn. Ximu Zhang, Email: zhangximu@hospital.cqmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement No datasets were generated or analysed during the current study.