An ultrasound-responsive hydrogel with piezoelectric-enhanced bioelectric-vesicular coupling alleviates mitochondrial dysfunction and inflammation for osteoarthritis therapy

Abstract

Osteoarthritis (OA) is a degenerative whole-joint disease characterized by interconnected pathological processes, including mitochondrial dysfunction, inflammatory microenvironment dysregulation, extracellular matrix (ECM) degradation, and aberrant subchondral bone remodeling. However, biological interventions or physical stimulation alone are often insufficient to coordinately regulate these processes. Here, we developed GM@KP-ABs, an injectable ultrasound-responsive hydrogel. It incorporates polydopamine-modified potassium sodium niobate nanoparticles (KNN@PDA) and Urolithin A-engineered apoptotic bodies (UA-ABs) within a gelatin methacryloyl matrix. KNN@PDA functioned as an electroactive transducer capable of converting low-intensity ultrasound into piezoelectric stimulation. Under these activation conditions, the treatment was associated with enhanced PINK1/Parkin-related mitophagic activity. In parallel, UA-ABs delivered complementary mitochondria-regulating signals through their vesicular cargo. The combined treatment was associated with enhanced clearance of damaged mitochondria and improved mitochondrial functional indices under osteoarthritic inflammatory conditions. These changes were accompanied by attenuated mitochondrial damage-associated activation of the cGAS-STING-NF-κB pathway, a treatment-associated shift of macrophages toward an anti-inflammatory M2-like phenotype, reduced inflammation, and improved ECM metabolic balance. In a destabilization of the medial meniscus (DMM)-induced OA model, ultrasound-activated GM@KP-ABs mitigated cartilage degeneration and aberrant subchondral bone remodeling. These findings support the potential of a remotely activated hydrogel platform that combines piezoelectric stimulation with engineered vesicular signaling for mitochondrial quality-control-oriented OA treatment.

Graphical abstract

Highlights

• •Injectable GM@KP-ABs couples piezoelectric and apoptotic-body signaling.
• •Ultrasound triggers electrical output and accelerates AB-associated protein release.
• •miR-122-5p cargo supports PINK1/Parkin-associated mitophagy.

1. Introduction

Osteoarthritis (OA) is a whole-joint degenerative disease characterized by cartilage degeneration, subchondral bone remodeling, synovial inflammation, and immune microenvironment deterioration [[1], [2], [3]]. Affecting more than 500 million people worldwide, OA represents a major clinical and socioeconomic burden [4]. Current treatments, including NSAIDs, intra-articular glucocorticoids, and sodium hyaluronate, mainly provide symptomatic relief but rarely prevent disease progression or restore joint function. Although joint replacement benefits end-stage OA, its invasiveness, limited lifespan, and long-term complications restrict its application, particularly in younger patients [5]. Therefore, disease-modifying strategies targeting OA pathogenesis are urgently needed.

Mitochondrial dysfunction is a key driver of OA progression [6,7]. Damaged mitochondria promote ROS accumulation, mitochondrial dysfunction, and ECM degradation, while mtDNA release activates cGAS-STING-NF-κB and inflammasome-related inflammatory pathways [[8], [9], [10], [11]]. Mitophagy selectively removes dysfunctional mitochondria, and impaired mitophagy perpetuates mitochondrial damage, inflammation, and ECM degradation [12,13]. Thus, enhancing mitochondrial quality control represents a promising OA therapeutic strategy [14]. However, OA involves interconnected pathological processes, including oxidative stress, inflammatory cascades, chondrocyte catabolism, macrophage polarization, and synovial inflammation, suggesting that multidimensional strategies coordinating biological and physical cues around mitochondrial regulation may provide improved therapeutic potential [[15], [16], [17], [18]].

Given the central role of mitochondrial quality control in OA progression, strategies capable of delivering mitochondria-regulating signals may provide therapeutic potential. Extracellular vesicle (EV)-based regulation has emerged as a promising approach for tissue repair and immune modulation [19,20]. Among EV subtypes, apoptotic bodies (ABs) are generated during apoptotic disassembly and retain diverse parental-cell components, including proteins, lipids, and nucleic acids [21]. This unique biogenesis enables ABs generation to be coupled with Urolithin A (UA) preconditioning, allowing incorporation of UA-induced mitochondrial regulatory cues. Moreover, phosphatidylserine exposure facilitates ABs uptake and supports inflammation resolution and tissue repair [22,23]. Since UA enhances mitophagy, improves mitochondrial function, and alleviates OA progression [24], UA-preconditioned BMSC-derived ABs were expected to exhibit mitochondrial regulatory activity. Consistently, UA-ABs showed enrichment of miR-122-5p, suggesting their potential role in mitochondrial quality control. However, free vesicles exhibit limited intra-articular retention and functional persistence due to degradation and joint motion [25]. Therefore, integrating engineered ABs with a biocompatible carrier and external physical stimulation may provide a more stable strategy for mitochondrial regulation in OA [26].

Physical bioelectrical regulation has emerged as an effective strategy for modulating cell behavior and tissue repair [[27], [28], [29]]. Piezoelectric biomaterials can generate localized electrical cues that regulate cellular electrophysiology, calcium signaling, mitochondrial function, and autophagy-related processes [30]. In cartilage repair, exercise-induced piezoelectric stimulation and biodegradable piezo-triboelectric platforms have been shown to promote cartilage regeneration in vivo [31]. In particular, ultrasound-responsive piezoelectric hydrogels enable non-invasive and remote electrical stimulation and have shown potential for cartilage regeneration and OA treatment [27,31,32]. Ultrasound-activated piezoelectric nanomaterials have also been reported to regulate mitophagy in inflammatory joint diseases, suggesting a link between bioelectrical cues and mitochondrial quality control [33]. Among these materials, lead-free potassium sodium niobate (KNN) exhibits favorable piezoelectric responsiveness and biosafety, making it attractive for biomedical applications [34,35]. However, existing piezoelectric hydrogel systems primarily provide physical stimulation with limited biochemical regulation. Conversely, EVs and ABs deliver bioactive cargos but lack externally controllable bioelectrical modulation. Therefore, integrating UA-engineered ABs with a KNN@PDA-based piezoelectric hydrogel (GM@KP-ABs) may provide complementary biological and physical regulation, providing a platform for investigating the combined regulation of mitochondrial quality-control-related and inflammatory processes in OA.

Collectively, these considerations provide a rationale for integrating UA-ABs with an ultrasound-activated piezoelectric hydrogel. UA-ABs provide mitochondria-regulating and immunomodulatory vesicular cargos, whereas ultrasound-activated KNN@PDA generates controllable piezoelectric bioelectrical cues (Scheme 1). These physical and vesicular inputs may both contribute to mitochondrial quality-control-related processes, including PINK1/Parkin-associated mitophagy [36,37], and may consequently be associated with increased damaged-mitochondrial clearance and attenuation of mtDNA-related inflammatory signaling. Therefore, GM@KP-ABs was designed as a bioelectric-vesicular platform to coordinately regulate mitochondrial dysfunction, inflammatory responses, ECM degradation, and OA progression. We investigated the physicochemical properties, ultrasound-responsive cellular responses, mitochondrial quality-control-related effects, inflammatory regulation, ECM protection, and therapeutic efficacy of GM@KP-ABs in vitro and in a DMM-induced OA model.

SCHEME 1.

UA-ABs and KNN@PDA were incorporated into GelMA and photocrosslinked under 405 nm light. Upon ultrasound stimulation, KNN@PDA-associated electromechanical cues and UA-AB-derived vesicular signals are proposed to contribute jointly to mitochondrial quality control and inflammatory regulation. Ultrasound-enhanced release of UA-AB-associated components may also contribute to the observed biological effects.

2. Results and discussion

2.1. Construction and characterization of KNN@PDA nanoparticles and UA-ABs

To construct an ultrasound-responsive composite hydrogel with both physical and biological regulatory functions, two functional modules were first prepared and characterized: KNN@PDA nanoparticles as the piezoelectric signal-generating unit and UA-ABs as the bioactive vesicular unit. Pristine KNN nanoparticles displayed an irregular block-like morphology (Fig. S1). After PDA surface modification, KNN@PDA largely retained the original particle morphology, while a thin low-electron-density layer was observed at the particle periphery, indicating the formation of a PDA coating on the KNN surface (Fig. 1A) [38]. EDS elemental mapping showed that Nb, K, Na, and O were distributed throughout the particles, whereas C and N signals were closely associated with the particle contour. Since nitrogen is absent from pristine KNN, the appearance of N signals provides direct evidence for PDA deposition. Fourier transform infrared spectroscopy (FTIR) further confirmed the introduction of PDA-related organic functional groups. Compared with pristine KNN, KNN@PDA exhibited a new absorption band at approximately 1291 cm⁻¹, which can be attributed to C-N stretching and/or phenolic C-O vibration associated with PDA (Fig. S2). Meanwhile, the main lattice vibration features of KNN were preserved, suggesting that PDA coating did not markedly alter the inorganic framework of KNN.

Fig. 1.

Construction and characterization of KNN@PDA nanoparticles and UA-ABs. (A) TEM image of KNN@PDA and the corresponding EDS elemental mapping; scale bar: 100 nm. (B) XRD patterns of KNN and KNN@PDA. (C) XPS survey spectra of KNN and KNN@PDA, showing enhanced C 1s and newly detected N 1s signals after PDA modification. (D) Live/Dead staining images of chondrocytes after different treatments; scale bar: 50 μm. (E) Quantification of live/dead cell percentages. (F) TEM image of UA-ABs, showing a spherical vesicular morphology; scale bar: 500 nm. (G) Annexin V-FITC/PI flow cytometry analysis of normal and apoptosis-induced BMSCs. (H) Flow cytometric analysis of the percentage of 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (Dil)-positive chondrocytes after treatment with Dil-labeled UA-ABs. (I) Annexin V-FITC/PI flow cytometric characterization of isolated UA-ABs. (J) Fluorescence imaging of time-dependent cellular uptake of Dil-labeled UA-ABs; scale bar: 50 μm. Sample size n = 3 for all quantitative experiments.

XRD analysis showed that KNN@PDA exhibited diffraction peaks similar to those of pristine KNN, with characteristic peaks at approximately 22^°, 32^°, 45^°, 46^°, 52^°, and 56^°, corresponding to the perovskite crystal structure of KNN (Fig. 1B). No additional impurity peaks or obvious peak shifts were observed after PDA modification, indicating that the crystalline structure of KNN was maintained during surface coating. XPS survey spectra further verified the change in surface chemical composition. Compared with bare KNN, KNN@PDA showed a markedly enhanced C 1s signal and a newly detectable N 1s signal, while the intrinsic KNN-related elements, including Nb, O, Na, and K, remained present (Fig. 1C). Collectively, the TEM, EDS, FTIR, XRD, and XPS results demonstrate that PDA was successfully introduced onto the KNN surface without disrupting the KNN crystalline framework.

Live/Dead staining was performed to evaluate the cytocompatibility of bare KNN and KNN@PDA under static and ultrasound-activated conditions. Bare KNN increased the number of dead chondrocytes and reduced the live-cell percentage compared with the control group. In contrast, PDA modification markedly improved cell viability, with the KNN@PDA group approaching the control level (Fig. 1D and E). This improvement may be attributed to PDA-mediated surface passivation, enhanced dispersion stability, and improved interfacial compatibility with cells [39]. Notably, ultrasound activation of KNN@PDA did not significantly increase cell death or reduce cell viability compared with KNN@PDA treatment alone. Consistently, in vivo histological assessment showed no detectable aggravation of joint tissue damage in the DMM + US group relative to the DMM group (Fig. S3). Collectively, these findings suggest that the ultrasound parameters used in this study did not induce measurable additional cytotoxicity or joint tissue injury under the tested experimental conditions. Nevertheless, low-intensity ultrasound may exert biological effects through mechanotransduction, inflammatory modulation, or changes in local mass transport. Therefore, the superior effects observed in the DMM + GM@KP-ABs + US group should be interpreted as the integrated outcome of ultrasound exposure, ultrasound-responsive KNN@PDA activation, and UA-AB-mediated biological regulation rather than as an effect attributable exclusively to piezoelectric stimulation.

To obtain bioactive vesicles with mitochondrial regulatory potential, BMSCs were preconditioned with UA before apoptosis induction. TEM showed that the resulting UA-ABs exhibited a spherical, membrane-enclosed morphology with submicron dimensions, consistent with the characteristics of apoptosis-derived extracellular vesicles [40]. Because UA-ABs were incorporated before GelMA photocrosslinking, their structural stability under 405 nm visible-light exposure was further examined. The illuminated UA-ABs retained a morphology comparable to that of the untreated UA-ABs, with no obvious membrane disruption, indicating that the applied photocrosslinking conditions did not appreciably compromise their structural integrity (Fig. 1F). In addition, Western blot analysis revealed a marked increase in apoptotic body-associated protein markers (Fig. S4), collectively confirming the successful generation of apoptotic body-like vesicles. Flow cytometry analysis showed that normal BMSCs were predominantly Annexin V⁻/PI⁻ viable cells, whereas apoptosis-induced BMSCs exhibited a marked increase in Annexin V-positive populations. The total Annexin V-positive ratio reached approximately 51.45%, indicating efficient induction of apoptosis and providing a sufficient cellular source for UA-ABs preparation (Fig. 1G).

To further verify the apoptotic origin of the isolated vesicles, UA-ABs were analyzed by Annexin V-FITC/PI staining. As shown in Fig. 1I, Annexin V-positive events were substantially enriched in the isolated UA-ABs, with a total Annexin V-positive ratio of approximately 94.52%. Because Annexin V binds to phosphatidylserine exposed on the outer leaflet of apoptotic membranes, this result supports the apoptosis-derived membrane phenotype of UA-ABs. The cellular uptake capability of UA-ABs was then evaluated using Dil labeling. Flow cytometry confirmed the efficient internalization of UA-ABs by chondrocytes (Fig. 1H). Fluorescence imaging demonstrated time-dependent uptake of Dil-labeled UA-ABs by chondrocytes. Intracellular red fluorescence progressively increased from 0 to 6 and 12 h and was predominantly distributed in the cytoplasm. UA-ABs exposed to 405 nm visible light exhibited a comparable uptake pattern, indicating that photocrosslinking illumination did not appreciably impair their cellular internalization (Fig. 1J). Moreover, light-exposed UA-ABs retained their ability to enhance colocalization between mitochondria and LC3B (Fig. S5). Together with the TEM observations, these results indicate that the applied photocrosslinking conditions preserved the structural integrity, cellular uptake, and mitophagy-related activity of UA-ABs.

Taken together, these results confirm the successful construction of two key functional modules for the composite hydrogel system. KNN@PDA served as a cytocompatible piezoelectric unit with preserved crystalline structure, while UA-ABs displayed apoptosis-derived vesicular features and efficient cellular uptake capability. These findings provide the material and biological basis for the subsequent fabrication of the GelMA/KNN@PDA/UA-ABs (GM@KP-ABs) ultrasound-responsive composite hydrogel.

2.2. Construction of ultrasound-responsive piezoelectric hydrogels and physicochemical and functional properties

Based on the functional modules established above, KNN@PDA nanoparticles and UA-ABs were incorporated into a GelMA matrix to construct an ultrasound-responsive composite hydrogel, denoted as GM@KP-ABs. GelMA was selected as the hydrogel matrix because of its photocrosslinkable network, high water content, injectability, and broad applicability in biomedical delivery systems. Hydrogel-based intra-articular delivery systems have also been widely explored to improve local retention and sustained release for OA treatment [41,42]. The precursor solution, containing GelMA, lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP), KNN@PDA, and UA-ABs, rapidly transformed into a stable hydrogel after 405 nm light irradiation (Fig. 2A). Before crosslinking, the precursor solution could be smoothly extruded through a syringe needle to form continuous patterns, demonstrating favorable injectability and in situ gelation potential for minimally invasive intra-articular administration.

Fig. 2.

Construction and physicochemical and functional characterization of the ultrasound-responsive piezoelectric hydrogel GM@KP-ABs. (A) Sol-gel transition of the GM@KP-ABs precursor after 405 nm visible-light-mediated photocrosslinking and injectability of the precursor solution; scale bar: 10 mm. (B) Scanning electron microscopy (SEM) image and energy-dispersive X-ray spectroscopy (EDS) elemental mapping of lyophilized GM@KP-ABs, showing an interconnected porous network and homogeneous elemental distribution; scale bar: 25 μm. (C) Swelling ratios of GM, GM@KP, and GM@KP-ABs hydrogels. (D) Compressive stress-strain curves, (E) compressive strength, and (F) Young's modulus of different hydrogels. (G) In vitro mass retention profiles of hydrogels over 28 days. (H) Frequency-dependent storage modulus (G′) and loss modulus (G″) of GM@KP-ABs. (I) Strain-dependent rheological behavior of GM@KP-ABs. (J) X-ray diffraction (XRD) patterns of GM, GM@KP, and GM@KP-ABs. (K) Piezoresponse force microscopy (PFM) amplitude-voltage curve of GM@KP-ABs. (L) Voltage output of GM@KP-ABs under periodic mechanical stimulation. (M) Ultrasound-triggered voltage output of GM@KP-ABs. (N) Cumulative ABs-associated protein release from GM@KP-ABs with or without ultrasound stimulation. (O) In vivo fluorescence imaging of Cy5-labeled GM@KP-ABs after intra-articular injection; scale bar: 10 mm. Sample size n = 3 for all quantitative experiments.

The microstructure of GM@KP-ABs was examined by SEM after freeze-drying. The hydrogel exhibited a continuous and interconnected porous network with pore sizes in the tens-of-micrometers range (Fig. 2B). Such a porous architecture is beneficial for water retention, molecular diffusion, and vesicle release within the hydrogel matrix [43]. EDS elemental mapping showed homogeneous distribution of C, N, and O from the organic hydrogel network and PDA-containing components, together with Nb, K, and Na signals derived from KNN@PDA nanoparticles. The uniform spatial distribution of these elements indicates that KNN@PDA was successfully incorporated into the GelMA network without obvious microscale aggregation, providing a structural basis for stable electromechanical functionality and local biological regulation.

The swelling behavior of the hydrogels was evaluated in phosphate-buffered saline (PBS) at 37 °C. GM, GM@KP, and GM@KP-ABs all exhibited high water absorption capacity, with equilibrium swelling ratios of approximately 1500-1600% and no statistically significant differences among the groups (Fig. 2C). These results indicate that the incorporation of KNN@PDA and UA-ABs did not markedly disrupt the primary GelMA network or its hydration capacity. For intra-articular application, this high water content is favorable for maintaining a hydrated soft microenvironment and facilitating molecular exchange between the hydrogel depot and surrounding tissues.

Compression testing further showed that all hydrogel groups exhibited typical soft hydrogel-like deformation behavior, with stress gradually increasing as compressive strain increased (Fig. 2D). Quantitative analysis showed that the compressive strength of GM@KP-ABs was approximately 14-15 kPa, while the Young's modulus was approximately 2 kPa, with no significant differences compared with GM and GM@KP (Fig. 2E and F). These results suggest that the introduction of KNN@PDA nanoparticles and UA-ABs did not compromise the soft mechanical characteristics or basic structural integrity of the GelMA hydrogel. Maintaining an appropriate soft modulus is important for reducing mechanical irritation and improving compatibility with the dynamic joint microenvironment.

The in vitro mass retention profiles were then examined in PBS at 37 °C. All hydrogels gradually lost mass over time, with approximately 20-30% of the initial mass remaining after 28 days (Fig. 2G). GM@KP-ABs showed a degradation profile comparable to those of GM and GM@KP, indicating that the incorporation of KNN@PDA and UA-ABs did not substantially alter the overall degradation behavior of the GelMA network. This gradual degradation profile is suitable for maintaining a local therapeutic window while reducing the risk of excessive long-term material retention.

The viscoelastic properties of GM@KP-ABs were characterized by rheological analysis. During frequency sweep testing, the storage modulus (G′) remained consistently higher than the loss modulus (G″), indicating that the hydrogel was dominated by elastic behavior and maintained a stable gel network over the tested frequency range (Fig. 2H). During strain sweep testing, G′ decreased progressively and eventually intersected with G″ as strain increased, reflecting network yielding and structural relaxation under large deformation (Fig. 2I). This rheological behavior suggests that GM@KP-ABs combines gel-like stability with deformation adaptability, which is relevant for mechanically dynamic joint environments. Supplementary rheological analysis further confirmed that both GM@KP and GM@KP-ABs retained typical gel characteristics. Compared with the other hydrogel formulations, GM@KP-ABs retained a similar network structure, indicating that the incorporation of KNN@PDA nanoparticles and UA-ABs did not disrupt the hydrogel network (Fig. S6).

To verify whether KNN@PDA retained its crystalline structure after hydrogel integration, XRD analysis was performed. GM mainly showed broad diffraction features associated with the amorphous hydrogel matrix, whereas GM@KP-ABs and GM@KP exhibited distinguishable KNN-related diffraction peaks at approximately 22° and 32° (Fig. 2J). These peaks are consistent with the characteristic reflections of KNN-based perovskite materials, indicating that KNN@PDA preserved its crystallographic features after incorporation into the GelMA hydrogel. This structural retention provides a material basis for subsequent piezoelectric functionality.

The local electromechanical behavior of GM@KP-ABs was evaluated by PFM. The amplitude-voltage curve displayed a butterfly-like profile under direct current bias scanning, indicating detectable bias-dependent local piezoresponse behavior within the composite hydrogel (Fig. 2K). Since PFM measurements were performed on dried hydrogel sections, this result should be interpreted as evidence of local electromechanical activity rather than a direct quantitative measure of piezoelectric output under hydrated physiological conditions. Macroscopic electrical output measurements further demonstrated that GM@KP-ABs generated reproducible pulsed voltage signals under periodic mechanical stimulation, with peak output reaching approximately ±2 V (Fig. 2L). These results confirm that KNN@PDA endowed the GelMA hydrogel with effective mechanical-to-electrical conversion capability.

Under ultrasound stimulation, hydrated GM@KP-ABs generated an oscillatory bulk voltage output of approximately 0.6-0.8 V (Fig. 2M), confirming ultrasound-induced electromechanical conversion at the construct level. However, this macroscopic voltage does not represent the local electric-field magnitude at the hydrogel-cell interface. Therefore, it provides material-level evidence of piezoelectric activation but should not be interpreted as a direct quantitative measurement of the electrical stimulus experienced by individual cells. A literature-informed ultrasound regimen of 1 MHz and 20 mW cm⁻² for 5 min every 2 days was selected as an experimental condition for activating the KNN@PDA-containing hydrogel. This regimen was not derived from a systematic dose-optimization study and should not be interpreted as an optimized therapeutic window [27]. Therefore, the enhanced biological effects observed in the GM@KP-ABs + US group were associated with ultrasound activation of the KNN@PDA-containing hydrogel together with UA-AB-mediated vesicular regulation. This ultrasound-triggered electrical output demonstrates that the composite hydrogel can be remotely activated by non-invasive mechanical stimulation, consistent with previous ultrasound-responsive piezoelectric hydrogel systems. Considering the intra-articular application scenario, such remote activation provides a practical route for delivering localized bioelectrical cues without implanted electrodes.

In addition to electrical activation, ultrasound stimulation also accelerated vesicle-associated protein release from the hydrogel network. In the absence of ultrasound, ABs-associated protein was continuously released from GM@KP-ABs over time, reaching a cumulative release of approximately 70-75% by day 20 (Fig. 2N). Periodic ultrasound stimulation increased the cumulative release to approximately 85-90%. This ultrasound-accelerated release may be attributed to microscale mechanical perturbation of the hydrogel network and enhanced mass transport under acoustic stimulation. Therefore, GM@KP-ABs provides not only ultrasound-triggered piezoelectric output but also ultrasound-enhanced bioactive vesicle release, demonstrating the coexistence of ultrasound-responsive electromechanical output and ultrasound-enhanced vesicle-associated release. This ultrasound-enhanced release represents an additional mechanism by which ultrasound may increase the biological effects of GM@KP-ABs. Therefore, differences between GM@KP-ABs and GM@KP-ABs + US cannot be interpreted solely as the contribution of piezoelectric activation.

The intra-articular retention behavior of the hydrogel system was further evaluated by in vivo fluorescence imaging. After intra-articular injection and in situ photocrosslinking, the fluorescence signal from Cy5-labeled GM@KP-ABs remained mainly localized in the knee joint region and was still detectable after 14 days (Fig. 2O). These results indicate that the GelMA-based composite hydrogel effectively prolonged local retention in the joint cavity, which is essential for sustained intervention in the OA microenvironment. In addition, H&E staining of major organs, including the heart, liver, spleen, lung, and kidney, showed no obvious pathological abnormalities after treatment with GM@KP-ABs or GM@KP-ABs + US, supporting the preliminary systemic biocompatibility of the hydrogel formulation (Fig. S7). Although joint and major-organ histology showed no overt abnormalities, these findings only support short-term tolerability. Real-time temperature changes, cumulative toxicity from prolonged ultrasound exposure, and the long-term degradation, biodistribution, and clearance of KNN@PDA remain unclear. Further thermometry, chronic toxicity, and elemental tracing studies in large-animal models are required before clinical translation.

Collectively, GM@KP-ABs exhibited favorable injectability, interconnected porous architecture, high swelling capacity, appropriate soft mechanical properties, gradual mass loss, stable viscoelastic behavior, preserved KNN crystallinity, local piezoresponse, ultrasound-triggered electrical output, ultrasound-accelerated ABs-associated protein release, prolonged intra-articular retention, and preliminary systemic biocompatibility. These properties establish GM@KP-ABs as an ultrasound-responsive bioactive piezoelectric hydrogel platform for subsequent mitochondrial regulation and OA therapeutic evaluation.

2.3. Ultrasound-responsive GM@KP-ABs reshapes the transcriptomic characteristics of chondrocytes under inflammatory stimulation

To clarify the transcriptional changes induced by GM@KP-ABs + US in inflammatory chondrocytes, RNA sequencing was performed on IL-1β-stimulated chondrocytes with or without GM@KP-ABs + US treatment. Differential expression analysis identified 1400 differentially expressed genes (DEGs), including 371 upregulated and 1029 downregulated genes, in the IL-1β + GM@KP-ABs + US group compared with the IL-1β group (Fig. 3A and B), using established RNA-seq differential expression workflows. As shown in the heatmap, genes associated with mitochondrial quality control and autophagy, including Pink1, Prkn, Optn, Atg5, and Atg12, were upregulated after GM@KP-ABs + US treatment (Fig. 3C). Cartilage matrix-associated anabolic genes, such as Col2a1 and Sox9, were also increased, suggesting a potential restoration of matrix-producing chondrocyte activity. In contrast, inflammation- and catabolism-related genes, including Il1b, Ptgs2, Nfkb1, Sting1, Mmp13, Mmp3, and Adamts5, were markedly downregulated. These results suggest that GM@KP-ABs + US treatment may simultaneously promote mitochondrial quality control, suppress inflammatory activation, and rebalance ECM metabolism in IL-1β-injured chondrocytes.

Fig. 3.

Ultrasound-activated GM@KP-ABs reshapes the transcriptomic landscape of IL-1β-stimulated chondrocytes. (A) Statistical summary of differentially expressed genes (DEGs) between the IL-1β group and IL-1β + GM@KP-ABs + US group. (B) Volcano plot showing upregulated and downregulated DEGs after GM@KP-ABs + US treatment. (C) Heatmap showing representative DEGs associated with mitophagy, ECM metabolism, and inflammatory signaling. (D) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEGs. (E) Reactome pathway enrichment analysis of DEGs. (F) Gene Set Enrichment Analysis (GSEA) of NF-κB signaling, mitophagy, and inflammatory response gene sets. Only the IL-1β + GM@KP-ABs + US group received ultrasound stimulation at 1 MHz and 20 mW cm⁻² for 5 min every 2 days. RNA sequencing was performed using three independently cultured chondrocyte samples per group (n = 3 biological replicates per group).

Pathway enrichment analyses were then performed to define the biological processes associated with these DEGs. KEGG enrichment analysis revealed significant enrichment in pathways related to cytokine-cytokine receptor interaction, ECM-receptor interaction, NF-κB signaling, and autophagy (Fig. 3D) [44]. Reactome enrichment analysis further highlighted the involvement of NF-κB activation, PI3K-Akt signaling, chemotaxis, cell migration, and stress-response-related pathways (Fig. 3E). Furthermore, GSEA based on the whole transcriptome provided additional pathway-level evidence. NF-κB signaling and inflammatory response gene sets showed negative enrichment after GM@KP-ABs + US treatment, whereas the mitophagy-related gene set showed positive enrichment (Fig. 3F) [45]. These results are consistent with the heatmap analysis and support a regulatory pattern characterized by enhanced mitochondrial quality control and attenuated inflammatory activation. Together, the transcriptomic data provide an unbiased molecular clue that ultrasound-activated GM@KP-ABs may coordinate mitophagy activation, inflammatory suppression, and ECM metabolic rebalancing, thereby providing the rationale for subsequent validation at the cellular and protein levels.

2.4. Ultrasound-responsive GM@KP-ABs is associated with improved mitochondrial function and increased PINK1/Parkin-related mitophagic activity

In this experimental design, GM@KP was used to evaluate the basal effect of the piezoelectric hydrogel matrix, GM@KP-ABs was used to assess the additional biological contribution of UA-ABs, and GM@KP-ABs + US was used to evaluate the combined effects observed when the KNN@PDA-containing hydrogel and UA-ABs were exposed to ultrasound; however, this comparison alone does not quantitatively distinguish electromechanical activation from ultrasound-enhanced UA-ABs release. To characterize cellular responses to ultrasound activation, DiBAC4(3)-based membrane-potential imaging and Fluo-4 AM-based intracellular Ca²⁺ imaging were performed (Fig. 4B, F, G). Compared with the ultrasound-alone and GM@KP-ABs-without-ultrasound groups, GM@KP-ABs + US significantly increased DiBAC4(3) fluorescence, consistent with chondrocyte membrane depolarization, and enhanced intracellular Ca²⁺ fluorescence. Both responses were attenuated by BAPTA-AM, a membrane-permeable intracellular Ca²⁺ chelator, supporting the involvement of intracellular Ca²⁺ signaling. Together with the material-level voltage-output measurements, these findings provide cellular-level evidence of membrane-potential-associated and Ca²⁺ responses under piezoelectric activation conditions. These observations are consistent with, but do not establish, a contribution of ultrasound-activated KNN@PDA to mitochondrial regulation. However, these fluorescence measurements neither quantify the local electric field nor fully exclude non-electrical ultrasound effects. Consistent with our findings, Gao et al. reported that Ca²⁺/CaMKII signaling activates Parkin and promotes PINK1/Parkin-mediated mitophagy [46], providing a plausible mechanistic link between the observed Ca²⁺ response and enhanced mitophagy (Fig. 4A). To further examine whether the intracellular Ca²⁺ response was involved in PINK1/Parkin-associated mitophagy regulation, chondrocytes were pretreated with BAPTA-AM before GM@KP-ABs + US treatment. Western blot analysis showed that BAPTA-AM significantly attenuated the increases in PINK1 and Parkin expression induced by GM@KP-ABs + US and reduced the LC3B-II/LC3B-I ratio (Fig. S8). These findings support the involvement of intracellular Ca²⁺ signaling in the treatment-associated mitophagy response. To distinguish the contributions of ultrasound, GelMA, KNN@PDA, and UA-ABs, different groups were evaluated using DiBAC4(3) staining for membrane potential changes and Fluo-4 AM staining for intracellular Ca²⁺ levels. Ultrasound alone and GM + US induced minimal changes in membrane potential and Ca²⁺ signals. In contrast, GM@KP + US showed stronger DiBAC4(3) and Fluo-4 fluorescence, indicating that the presence of KNN@PDA under ultrasound stimulation was associated with stronger membrane-potential-associated and intracellular Ca²⁺ responses. GM@KP-ABs + US showed comparable membrane potential and Ca²⁺ responses to GM@KP + US, with no significant difference between the two groups. These findings suggest that UA-ABs mainly contribute to bioactive regulation, whereas the presence of KNN@PDA under ultrasound stimulation was associated with stronger electrophysiological responses (Fig. S9). The present controls distinguish the KNN@PDA-associated contribution to membrane-potential-related and Ca²⁺ responses but do not provide equivalent component-resolved comparisons for downstream mitochondrial and inflammatory endpoints. Thus, the enhanced PINK1/Parkin-associated mitophagy and inflammatory suppression observed after GM@KP-ABs + US treatment may result from the combined effects of electromechanical activation, accelerated UA-AB release, and vesicular bioactivity.

Fig. 4.

Regulatory effects of ultrasound-responsive GM@KP-ABs on mitochondrial function and PINK1/Parkin-mediated mitophagy. (A) Schematic illustration of the mechanism by which ultrasound-responsive GM@KP-ABs regulates PINK1/Parkin-mediated mitophagy. (B, F, G) DiBAC4(3)-based membrane-potential-associated fluorescence and Fluo-4 AM-based intracellular Ca²⁺ fluorescence; scale bars: 20 μm and 200 μm, respectively. (C, H) JC-1 staining and quantitative analysis of mitochondrial membrane potential; scale bar: 20 μm. (D, I) MitoSOX staining for mitochondrial superoxide and MitoTracker staining for mitochondrial morphology; scale bar: 20 μm. (E, J) mRFP-GFP-LC3 tandem fluorescence assay and quantitative analysis of general autophagic flux; scale bar: 5 μm. (K-O) Western blot analysis and quantification of PINK1, Parkin, LC3B, and p62. Data are presented as mean ± SD. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3.

Damaged mitochondria can release mtDNA and other danger-associated molecular signals into the cytosol, thereby activating cGAS-STING-NF-κB-dependent sterile inflammatory signaling [47]. Thus, enhanced mitophagy clears dysfunctional mitochondria, lowering their inflammatory output and thereby dampening the cGAS-STING-NF-κB axis [48]. Building on transcriptomic evidence that suggests a role for mitophagy and inflammation regulation, we explored whether GM@KP-ABs + US treatment could improve mitochondrial function and mitochondrial quality-control-related features in IL-1β-stimulated chondrocytes. Unless otherwise stated, only the GM@KP-ABs + US group received ultrasound stimulation at 1 MHz and 20 mW cm⁻² for 5 min every 2 days. Our findings revealed that IL-1β significantly induced mitochondrial oxidative stress and morphological damage, but treatment with GM@KP-ABs + US effectively reversed these issues. MitoSOX staining showed that IL-1β caused a substantial accumulation of mitochondrial superoxide, which was reduced by both the GM@KP-ABs group and the GM@KP-ABs + US group, with the latter having the most pronounced inhibitory effect (Fig. 4D). MitoTracker staining further demonstrated that IL-1β disrupted mitochondrial continuity, leading to fragmentation. However, GM@KP-ABs + US treatment improved mitochondrial network continuity and improved several mitochondrial morphological parameters (Fig. 4D and Fig. S10). Additionally, a similar trend was observed in intracellular reactive oxygen species (ROS) levels measured with the 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) probe: ROS levels increased significantly after IL-1β stimulation and decreased substantially following GM@KP-ABs + US treatment (Fig. S11). Overall, these results suggest that GM@KP-ABs + US treatment effectively mitigates IL-1β-induced mitochondrial oxidative stress, fragmentation, and loss of mitochondrial content, thereby improving mitochondrial function and quality control in chondrocytes. Mitochondrial membrane potential was then evaluated using JC-1 staining. IL-1β stimulation caused a pronounced decrease in the red/green fluorescence ratio, indicating mitochondrial membrane potential dissipation and mitochondrial dysfunction (Fig. 4C, H). Treatment with GM@KP, GM@KP-ABs, and GM@KP-ABs + US partially recovered mitochondrial membrane potential to varying degrees, with the GM@KP-ABs + US group producing the most significant recovery. This result demonstrates that ultrasound-responsive GM@KP-ABs not only reduced mitochondrial oxidative stress but also improved mitochondrial functional integrity under inflammatory conditions. This apparent coexistence of improved mitochondrial membrane potential and enhanced PINK1/Parkin-related mitophagy suggests that GM@KP-ABs + US treatment did not induce global mitochondrial depolarization. Instead, the treatment may facilitate the selective recognition and removal of dysfunctional mitochondria, thereby improving the functional status of the overall mitochondrial population.

To evaluate general autophagic flux after treatment, the mRFP-GFP-LC3 tandem fluorescence reporter was used. Red-only puncta represent autolysosomal structures, whereas yellow puncta represent autophagosomal structures [49]. In this study, mitochondrial quality control refers specifically to oxidative stress regulation, membrane-potential maintenance, mitophagy-associated clearance, and mtDNA-related inflammatory signaling. Compared with the IL-1β group, the GM@KP-ABs group and the GM@KP-ABs + US group significantly increased the number of autolysosomes, with the highest level observed in the GM@KP-ABs + US group (Fig. 4E, J). This finding is consistent with increased general autophagic flux under the tested conditions rather than the accumulation of autophagosomes alone; however, it does not establish mitochondria-specific mitophagy flux. To further assess mitochondrial-selective autophagy, LC3B-mitochondria and TOMM20-lysosome colocalization analyses were performed. GM@KP-ABs + US treatment markedly increased LC3B colocalization with mitochondria and enhanced TOMM20 colocalization with lysosomes, indicating promoted recruitment of autophagic machinery to mitochondria and subsequent mitochondrial delivery to lysosomes (Figs. S12 and S13). These colocalization results support increased recruitment of autophagic machinery to mitochondria and increased delivery of mitochondrial components to lysosomal compartments after GM@KP-ABs + US treatment.

Western blot analysis further showed changes in mitophagy-associated proteins consistent with enhanced mitophagy-related activity. IL-1β suppressed PINK1, Parkin, and LC3B-II expression while increasing p62 accumulation, indicating impaired mitochondrial quality control and blocked autophagic degradation. GM@KP-ABs + US treatment partially reversed these changes, as shown by the upregulation of PINK1, Parkin, and LC3B-II, together with reduced p62 levels (Fig. 4K–O). As the reduction in p62 coincides with the rise in autolysosome formation, these findings are consistent with increased autophagic processing and enhanced mitophagy-related activity, but do not directly quantify mitochondria-specific autophagic flux. Although the mRFP-GFP-LC3 reporter demonstrated increased general autophagic flux, it was not mitochondria-specific. The mitochondrial colocalization and Parkin-intervention experiments provided complementary spatial and functional evidence of mitophagy-related activity; however, a mitochondria-targeted flux reporter, such as mt-Keima or mito-QC, would be required for direct dynamic quantification of mitophagy flux.

Taken together, these results indicate that ultrasound-responsive GM@KP-ABs + US treatment was associated with improvements in the measured mitochondrial functional indices and increases in mitophagy-related activity. UA-ABs contributed vesicular signals related to mitochondrial protection and mitophagy, while the KNN@PDA-containing hydrogel was associated with stronger responses under ultrasound activation. These differences may reflect the combined contributions of electromechanical activation, ultrasound-enhanced UA-AB-associated release, and vesicular bioactivity. Importantly, the partial recovery of mitochondrial membrane potential and activation of PINK1/Parkin-related mitophagy suggest that GM@KP-ABs + US treatment did not induce nonspecific mitochondrial depolarization. Instead, the system appears to promote the recognition and clearance of dysfunctional mitochondria while improving the measured functional indices of the overall mitochondrial population. This interpretation is consistent with the reduced mitochondrial superoxide, improved mitochondrial morphology, partially recovered mitochondrial membrane potential, enhanced autophagic flux, and increased mitochondria-lysosome colocalization observed after GM@KP-ABs + US treatment. These analyses primarily addressed mitochondrial oxidative stress, membrane-potential abnormalities, mitophagy-associated activity, and mtDNA-related inflammatory signaling. However, mitochondrial fusion/fission dynamics and mitochondrial biogenesis were not directly assessed in the present study.

Mechanistically, PINK1/Parkin-mediated mitophagy is a central mitochondrial quality-control pathway involved in the recognition and autophagic clearance of damaged mitochondria [50]. In the present GM@KP-ABs + US system, material-level measurements confirmed the electromechanical conversion capability of KNN@PDA, while cellular experiments showed membrane-potential-associated and Ca²⁺ responses under ultrasound stimulation. Together with the observed changes in mitophagy-related proteins, autophagic flux, and mitochondria-lysosome colocalization, these findings are consistent with a contribution of ultrasound-responsive piezoelectric stimulation to mitochondrial quality control. UA-ABs provided complementary vesicular signals, and the combined treatment was associated with enhanced mitophagy-related activity, reduced mitochondrial ROS accumulation, and improved mitochondrial functional indices. However, because the local electric field was not quantified and non-electrical ultrasound effects were not fully excluded, the present data do not establish piezoelectric electrical output as the direct cause of mitophagy activation. The findings therefore support a model in which piezoelectric and vesicular inputs may jointly contribute to mitochondrial regulation and downstream inflammatory suppression.

2.5. Ultrasound-responsive GM@KP-ABs suppresses cGAS-STING-NF-κB signaling and remodels the inflammatory microenvironment

Building on the mitochondrial protective effects described above, we next investigated whether GM@KP-ABs + US treatment could suppress downstream inflammatory signaling in IL-1β-stimulated chondrocytes. Mitochondrial dysfunction and defective mitophagy are closely associated with sterile inflammatory activation, partly through the cGAS-STING-NF-κB pathway. In OA, activation of cGAS-STING and nuclear factor kappa B (NF-κB) signaling contributes to chondrocyte inflammatory activation, senescence, apoptosis, and ECM degradation [51]. Therefore, we examined whether the observed improvement in mitochondrial quality-control-related indices by ultrasound-responsive GM@KP-ABs was accompanied by inhibition of the cGAS-STING-NF-κB inflammatory axis.

Because impaired mitochondrial quality control can promote cytosolic mtDNA accumulation, we first examined the spatial association between mtDNA and cGAS. Immunofluorescence analysis revealed distinct mtDNA-cGAS colocalization patterns across the Control, IL-1β, GM@KP-ABs + US, GM@KP-ABs + US + si-PINK1, and GM@KP-ABs + US + si-PINK1 + VBIT-4 groups (Fig. 5A). PINK1 knockdown increased mtDNA-cGAS colocalization, whereas this effect was attenuated by VBIT-4, an inhibitor of voltage-dependent anion channel 1 (VDAC1) oligomerization that limits mitochondrial outer membrane permeabilization and mtDNA release. These results suggest that disruption of PINK1-associated mitochondrial quality control promotes mtDNA release and its subsequent association with cGAS.

Fig. 5.

Ultrasound-responsive GM@KP-ABs suppresses cGAS-STING-NF-κB signaling, reducing inflammatory factor release and promoting M2 macrophage polarization to reshape the inflammatory microenvironment. (A) Immunofluorescence analysis of spatial colocalization between mitochondrial DNA (mtDNA, green) and cGAS (red) in chondrocytes; scale bar: 20 μm. (B, C) Immunofluorescence staining and quantitative analysis show the expression level of COX-2. Scale bar: 50 μm. (D, E) Immunofluorescence staining and quantitative analysis show the expression level of iNOS. Scale bar: 50 μm. (F, G) Western blot and quantitative analysis show the expression changes of cGAS, STING, and p-p65/p65. (H, I) Immunofluorescence staining and quantitative analysis of p65 nuclear localization in chondrocytes; scale bar: 20 μm. (J, K) Western blot and quantitative analysis show the expression changes of macrophage polarization-related marker proteins CD206, CD86, and ARG1 in RAW264.7 cells. (L) Quantification of inflammatory mediators and cytokines, including iNOS, COX-2, TNF-α, IL-1β, and IL-6. Error bars show mean ± SD. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3.

Cytosolic mtDNA levels were quantified by qPCR using four mitochondrial genome markers (Nd1, Cytb, Nd4, and Co1). Primer sequences for these markers are listed in Table S1. All four markers showed a comparable trend: IL-1β significantly increased cytosolic mtDNA abundance [52], indicating enhanced mitochondrial membrane disruption and mtDNA leakage, whereas incorporation of ABs into GM@KP gradually decreased mtDNA accumulation, with the most pronounced reduction observed after ultrasound activation of GM@KP-ABs (Figs. S14 and S15). These findings provide molecular evidence that GM@KP-ABs + US was associated with improved mitochondrial quality-control-related indices and limited the release of mitochondrial danger signals upstream of cGAS-STING inflammatory signaling.

Western blot analysis showed that IL-1β markedly increased cGAS and STING expression and promoted NF-κB p65 phosphorylation, as reflected by an elevated p-p65/p65 ratio (Fig. 5F–G). GM@KP treatment showed limited inhibitory effects, whereas GM@KP-ABs reduced these inflammatory signaling markers to a greater extent. Notably, GM@KP-ABs + US treatment produced the strongest suppression of cGAS, STING, and p-p65/p65, indicating that the combined treatment was associated with attenuation of the cGAS-STING-NF-κB axis in inflamed chondrocytes. To further evaluate downstream NF-κB activation, the nuclear translocation of p65 was examined by immunofluorescence staining. DMXAA (5,6-dimethylxanthenone-4-acetic acid), a pharmacological agonist that directly activates murine STING, markedly increased nuclear p65 accumulation. In contrast, GM@KP-ABs + US treatment and STING silencing both reduced DMXAA-induced p65 nuclear localization and the expression of iNOS, COX-2, TNF-α, IL-1β, and IL-6. These effects are shown in Fig. 5H–I and Fig. S16. These results further support the ability of GM@KP-ABs + US to suppress STING-associated NF-κB activation.

We then assessed downstream inflammatory mediators. Immunofluorescence staining showed that IL-1β stimulation strongly upregulated COX-2 and iNOS expression, while GM@KP-ABs + US treatment markedly reduced both signals (Fig. 5B–E). Consistently, Western blot analysis further confirmed the downregulation of COX-2 and iNOS after GM@KP-ABs treatment (Fig. S17). Quantitative analysis of inflammatory mediators and cytokines showed that GM@KP-ABs + US treatment significantly decreased iNOS, COX-2, TNF-α, IL-1β, and IL-6 levels compared with the IL-1β group (Fig. 5L). These results indicate that ultrasound-responsive GM@KP-ABs effectively suppresses IL-1β-induced inflammatory factor release in chondrocytes. In addition, flow cytometric apoptosis analysis showed that ultrasound-responsive GM@KP-ABs reduced IL-1β-induced chondrocyte apoptosis, further indicating that the composite hydrogel protected chondrocytes from inflammatory injury (Fig. S18).

Given that the inflammatory microenvironment of osteoarthritis is shaped not only by chondrocytes but also by synovial immune cells [53], we further evaluated the effect of GM@KP-ABs + US treatment on macrophage polarization. Macrophage phenotypic switching is closely associated with OA progression, and synovial immune cells can influence disease progression by amplifying or alleviating local inflammatory responses. M1-like macrophages generally promote inflammatory cytokine production, whereas M2-like macrophages are associated with inflammation resolution and tissue repair, although macrophage polarization should be interpreted as a spectrum rather than a fixed binary state [54]. Western blot analysis showed that GM@KP-ABs + US treatment significantly increased the expression of M2-associated markers CD206 and ARG1 while reducing the M1-associated marker CD86 (Fig. 5J–K). Compared with the GM@KP group and the GM@KP-ABs group, the GM@KP-ABs + US group exerted a more pronounced shift toward an anti-inflammatory M2-like phenotype, suggesting that the composite hydrogel can modulate macrophage polarization in addition to suppressing inflammatory signaling in chondrocytes. To determine whether GM@KP-ABs + US directly modulates macrophage phenotype, RAW264.7 cells were treated with the indicated formulations in the absence of chondrocytes. CD206 immunofluorescence remained relatively low in the IL-1β and GM@KP groups, whereas GM@KP-ABs moderately increased CD206 expression. Notably, GM@KP-ABs + US produced the strongest CD206 fluorescence, with a significant increase compared with all other groups (Fig. S19). Together with the reduced CD86 and increased ARG1 expression observed by Western blot, these findings support an association between the treatment and a shift of RAW264.7 cells toward a CD206-high, M2-like phenotype.

The in vitro macrophage polarization results were further supported by in vivo synovial tissue staining. In the destabilization of the medial meniscus (DMM)-induced OA model, synovial immunohistochemistry showed lower CD86 and higher CD206 expression in the GM@KP-ABs + US group than in the DMM group (Fig. S20). Because these markers were not colocalized with a macrophage-specific marker in this experiment, these single-marker changes alone were not interpreted as cell-specific evidence of an M1-to-M2 transition. F4/80/CD206 co-staining further showed increased F4/80+CD206+ double-positive macrophages following treatment, with the most pronounced increase in the GM@KP-ABs + US group (Fig. S21). These findings support a treatment-associated shift of synovial macrophages toward an M2-like phenotype under the tested conditions but do not establish a reduction in M1 macrophage abundance.

Overall, ultrasound-responsive GM@KP-ABs suppressed the cGAS-STING-NF-κB inflammatory axis in chondrocytes, reduced inflammatory mediator production, protected against IL-1β-induced apoptotic injury, and promoted a treatment-associated shift of macrophages toward an anti-inflammatory phenotype. Compared with GM@KP-ABs without ultrasound activation, the GM@KP-ABs + US group produced more pronounced anti-inflammatory effects. This difference may reflect the combined contributions of ultrasound exposure, KNN@PDA-associated electromechanical activation, and ultrasound-enhanced UA-AB-associated release. These effects were consistent with the improvements in measured mitochondrial quality-control-related features described above. Together, the data support an association between treatment-related mitochondrial regulation and inflammatory microenvironment remodeling, while the relative contributions of the physical and vesicular components remain unresolved.

2.6. Ultrasound-responsive GM@KP-ABs treatment improves ECM-related anabolic-catabolic balance in IL-1β-stimulated chondrocytes

Cartilage ECM homeostasis depends on the balance between anabolic matrix components, such as Collagen II and Aggrecan, and catabolic enzymes that degrade collagen and proteoglycan networks [55]. After demonstrating that ultrasound-responsive GM@KP-ABs enhanced mitophagy-associated responses and suppressed inflammatory signaling, we further evaluated whether these effects translated into improved ECM metabolism in IL-1β-stimulated chondrocytes. Immunofluorescence staining showed that IL-1β markedly reduced the expression of Collagen II and Aggrecan, two key ECM components essential for maintaining cartilage structural integrity and proteoglycan-rich matrix organization. Notably, GM@KP-ABs + US treatment produced the greatest increases in Collagen II and Aggrecan fluorescence relative to the IL-1β group, indicating enhanced matrix-anabolic activity under inflammatory conditions (Fig. 6A–D).

Fig. 6.

Ultrasound-responsive GM@KP-ABs improves ECM anabolic-catabolic balance in IL-1β-stimulated chondrocytes. (A, B) Immunofluorescence staining and quantitative analysis of Collagen II expression; scale bar: 50 μm. (C, D) Immunofluorescence staining and quantitative analysis of Aggrecan expression; scale bar: 50 μm. (E, F) Immunofluorescence staining and quantitative analysis of MMP13 expression; scale bar: 50 μm. (G, H) Western blot analysis and corresponding quantification of Aggrecan, Collagen II, and MMP13 expression. Data are presented as mean ± SD. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3.

Matrix catabolism in OA is largely driven by proteolytic enzymes that degrade the collagen-rich cartilage framework. Among these enzymes, MMP13 is a major collagenase involved in collagen II degradation [56]. We therefore examined MMP13 as a representative matrix-catabolic marker. IL-1β markedly increased MMP13 expression, indicating activation of matrix-degradative responses in chondrocytes (Fig. 6E and F). Both GM@KP-ABs treatment and GM@KP-ABs + US treatment significantly reduced MMP13 fluorescence intensity compared with the IL-1β group, suggesting effective suppression of IL-1β-induced matrix catabolism. Although the difference between the GM@KP-ABs group and the GM@KP-ABs + US group was not significant in the immunofluorescence analysis, GM@KP-ABs + US treatment maintained a consistently lower MMP13 level, supporting its overall role in attenuating ECM degradation.

Western blot analysis further confirmed these immunofluorescence findings at the protein level. IL-1β decreased Aggrecan and Collagen II while increasing MMP13 expression. GM@KP-ABs + US treatment significantly upregulated Aggrecan and Collagen II and downregulated MMP13 compared with the IL-1β group (Fig. 6G and H). The greater increases in anabolic matrix proteins and suppression of MMP13 in the GM@KP-ABs + US group indicate that the composite hydrogel effectively rebalanced ECM anabolism and catabolism in inflamed chondrocytes.

Taken together, these results demonstrate that GM@KP-ABs + US treatment not only improved mitochondrial function and reduced inflammatory activation but also effectively improved ECM metabolic balance in IL-1β-injured chondrocytes. Notably, compared with the GM@KP-ABs group (without ultrasound activation), the GM@KP-ABs + US group produced a more pronounced recovery of matrix-anabolic markers and a stronger suppression of MMP13-mediated catabolic activity. Because ultrasound simultaneously activates KNN@PDA-associated electromechanical responses and enhances UA-AB-associated release, this difference should be interpreted as a combined treatment effect rather than a piezoelectric-specific effect. These findings provide cellular-level evidence that the biological regulation mediated by UA-ABs and the physical bioelectrical cues generated by ultrasound-activated KNN@PDA may provide complementary contributions to preserve cartilage matrix metabolism under inflammatory OA-like conditions.

2.7. miR-122-5p is involved in UA-ABs-mediated mitophagy activation and inflammatory suppression

Previous piezoelectric hydrogels promote cartilage repair through ultrasound-generated electrical cues but generally lack sustained biochemical regulation, whereas extracellular-vesicle and apoptotic-body therapies deliver bioactive cargos but lack externally controlled bioelectric stimulation. GM@KP-ABs integrates KNN@PDA and UA-engineered apoptotic bodies within an injectable GelMA depot, combining ultrasound-triggered bioelectric cues with miR-122-5p-enriched vesicular cargo. These physical and vesicular inputs may both contribute to PINK1/Parkin-associated mitochondrial quality-control-related processes, promoting damaged mitochondrial clearance and suppressing mtDNA-mediated cGAS-STING-NF-κB activation. Thus, the principal innovation lies in the coupled bioelectric-vesicular regulation of mitochondrial quality control rather than in any individual component.

To clarify the molecular basis of UA-AB bioactivity, the miRNA profiles of UA-ABs and staurosporine-induced apoptotic bodies (STS-ABs) were compared by small RNA sequencing. UA-ABs exhibited a distinct miRNA cargo profile, with 91 upregulated and 57 downregulated miRNAs relative to STS-ABs (Fig. 7A, B, E). Among them, miR-122-5p was markedly enriched in UA-ABs, identifying it as a potential functional cargo involved in UA-AB-mediated mitochondrial regulation. Bioinformatic analysis predicted a putative miR-122-5p recognition site within the TGFBR1 3′ untranslated region (Fig. 7C). Consistent with a possible pathway-level association, a previous study showed that cVIM-mediated sequestration of miR-122-5p and miR-9-5p was associated with increased TGFBR1/TGFBR2 expression and TGF-β/Smad activation, whereas exogenous miR-122-5p attenuated this signaling cascade [57]. However, direct binding between miR-122-5p and the TGFBR1 3’ untranslated region was not examined using a dual-luciferase reporter assay, and the functional involvement of TGFBR1 was not evaluated using knockdown, overexpression, or rescue experiments. Therefore, TGFBR1 should be regarded as a predicted downstream candidate rather than a validated direct target. Pathway enrichment analysis of the predicted miRNA targets further identified TGF-β signaling- and autophagy-related pathways, supporting the potential involvement of UA-AB-enriched miRNAs in mitochondrial quality control and inflammatory regulation (Fig. 7D).

Fig. 7.

miR-122-5p contributes to UA-ABs-induced activation of mitophagy and suppression of inflammatory signaling. (A, B) Differential expression analysis and statistical summary of miRNAs in UA-ABs and STS-ABs. (C) Bioinformatic prediction of a putative regulatory relationship between miR-122-5p and the TGFBR1 3′ untranslated region, interpreted together with published evidence regarding miR-122-5p-associated TGF-β signaling. (D) Pathway enrichment analysis based on the predicted target genes of differentially expressed miRNAs. (E) Heatmap showing representative differentially expressed miRNAs between UA-ABs and STS-ABs. (F, G) Western blot analysis and quantification of mitophagy-related proteins, including PINK1, Parkin, LC3B, and p62, and inflammatory signaling-related proteins, including cGAS, STING, and p-p65/p65, after Parkin knockdown or Mdivi-1-mediated inhibition. (H, I) Western blot analysis and quantification of the mitophagy-related proteins PINK1 and Parkin and the inflammatory signaling-related proteins cGAS, STING, and p-p65/p65 following lentivirus-mediated miR-122-5p overexpression in IL-1β-stimulated chondrocytes. Data are presented as mean ± SD. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3.

Because transcriptomic and cellular analyses suggested that GM@KP-ABs enhanced mitophagy while suppressing inflammatory signaling, we next examined whether mitophagy was functionally linked to inhibition of the cGAS-STING-NF-κB axis. Parkin knockdown and Mdivi-1 treatment were used to interfere with GM@KP-ABs-induced mitophagy. Western blot analysis showed that Parkin knockdown reduced Parkin expression and weakened the GM@KP-ABs-induced upregulation of mitophagy-related markers. Similarly, Mdivi-1 treatment attenuated the mitophagy-related regulatory effects of GM@KP-ABs, as reflected by decreased PINK1/Parkin-associated signaling, reduced LC3B levels, and partial restoration of p62 accumulation (Fig. 7F and G). Importantly, suppression of mitophagy was accompanied by reactivation of inflammatory signaling, as indicated by increased cGAS, STING, and p-p65/p65 levels. These results suggest that mitophagy activation is an important upstream event through which GM@KP-ABs suppresses the cGAS-STING-NF-κB inflammatory pathway.

To further establish the functional role of Parkin, an shRNA-resistant Parkin vector was introduced into Parkin-silenced chondrocytes. GM@KP-ABs + US increased the LC3B-II/LC3B-I ratio and reduced p62 expression, whereas Parkin knockdown reversed these effects. Parkin re-expression restored the mitophagy-associated protein profile and reduced the elevated p-p65/p65 ratio caused by Parkin silencing (Fig. S22). Consistently, Parkin knockdown attenuated the GM@KP-ABs + US-induced recovery of mitochondrial membrane potential, whereas Parkin re-expression partially restored mitochondrial membrane potential, as indicated by the JC-1 red-to-green fluorescence intensity ratio (Fig. S23).

Parkin knockdown also reduced Collagen II fluorescence and increased MMP13 mRNA expression, thereby weakening the ECM-protective effects of GM@KP-ABs + US. Re-expression of Parkin partially restored Collagen II fluorescence and significantly reduced MMP13 expression, although the increase in Collagen II mRNA did not reach statistical significance (Figs. S24 and S25). Collectively, these findings support a functional contribution of Parkin to the mitophagy-associated, anti-inflammatory, and ECM-protective effects of GM@KP-ABs + US.

To further determine whether miR-122-5p participates in the regulatory effect of GM@KP-ABs, miR-122-5p antagomir was introduced into the treatment system. Compared with GM@KP-ABs alone, miR-122-5p inhibition markedly reduced the colocalization of LC3B with mitochondria, indicating impaired formation of mitophagosomes (Fig. S26). Consistently, Western blot analysis showed that miR-122-5p inhibition decreased the expression of PINK1 and Parkin, while increasing cGAS and STING levels (Fig. S27). These results indicate that miR-122-5p contributes to GM@KP-ABs-mediated mitophagy activation and inflammatory pathway suppression. In addition, inhibition of miR-122-5p reduced Collagen II expression, increased MMP13 expression, and attenuated Aggrecan restoration, suggesting that miR-122-5p is also involved in the ECM-protective effects of GM@KP-ABs (Figs. S27 and S28). Conversely, lentiviral overexpression of miR-122-5p enhanced the PINK1/Parkin-associated mitophagy-related response and suppressed cGAS, STING, and p-p65/p65 signaling (Fig. 7H and I). In addition, miR-122-5p overexpression significantly reduced the IL-1β-induced expression of iNOS, COX-2, TNF-α, IL-1β, and IL-6 in chondrocytes (Fig. S29). These findings further support the involvement of miR-122-5p in mitophagy-related regulation and inflammatory suppression.

Taken together, these findings identify miR-122-5p as a functional miRNA cargo enriched in UA-ABs and suggest its involvement in the biological regulation mediated by GM@KP-ABs. Rather than acting through a single linear pathway, miR-122-5p appears to contribute to a coordinated regulatory network involving mitophagy activation, suppression of cGAS-STING-NF-κB signaling, and improvement of ECM anabolic-catabolic balance. Based on these results, UA-ABs-delivered miR-122-5p may promote mitochondrial quality control, thereby limiting damaged mitochondria-associated inflammatory signaling in IL-1β-stimulated chondrocytes. This biological effect may contribute alongside ultrasound-associated electromechanical activation and ultrasound-enhanced UA-AB release, suggesting complementary biological and physical contributions to the observed mitochondrial and inflammatory responses. These findings suggest that miR-122-5p is an important UA-ABs-enriched miRNA cargo that contributes to mitophagy activation, inflammatory signaling suppression, and ECM protection. Together, the bidirectional gain- and loss-of-function results support miR-122-5p as a functionally relevant UA-AB-enriched miRNA cargo associated with mitophagy-related regulation, inflammatory signaling suppression, and ECM protection. Although the present gain- and loss-of-function experiments establish the functional relevance of miR-122-5p, the downstream molecular targets responsible for these effects remain to be fully defined. In particular, direct regulation of TGFBR1 will require validation using wild-type and recognition-site-mutant TGFBR1 3’ UTR reporter constructs.

2.8. Ultrasound-responsive GM@KP-ABs attenuates OA progression and promotes cartilage and subchondral bone repair in vivo

To evaluate the in vivo therapeutic efficacy of GM@KP-ABs + US treatment, a DMM-induced mouse OA model was established, followed by intra-articular administration of different hydrogel formulations and ultrasound intervention (Fig. 8C). Previous studies support the transarticular activation of injectable piezoelectric hydrogels. Vinikoor et al. applied ultrasound at 40 kHz and 0.33 W cm⁻² to rabbit knee joints, promoting cartilage regeneration without evident inflammation or apoptosis [27]. LIPUS protocols for OA generally use frequencies of 1-3 MHz and intensities of approximately 0.02-1 W cm⁻² [58], with repeated exposure at comparable or higher intensities evaluated in clinical and preclinical studies [[59], [60], [61]]. Based on this evidence, we applied a low-intensity regimen of 1 MHz and 0.02 W cm⁻² for 5 min to activate GM@KP-ABs both in vitro and in vivo. The relatively low acoustic intensity and short exposure duration were selected to minimize tissue heating. The temperature of the coupling medium was monitored during the material-level electrical-output measurements to reduce potential thermal artifacts. However, this measurement does not represent the intra-articular temperature during in vivo treatment.

Fig. 8.

Ultrasound-responsive GM@KP-ABs attenuates osteoarthritis progression and alleviates cartilage degeneration and abnormal subchondral bone remodeling in vivo. (A) H&E staining and Safranin O/Fast Green staining show the cartilage structure and proteoglycan distribution in different treatment groups, scale bar: 200 μm. (B) The OARSI score was used to quantitatively assess cartilage degeneration by evaluators blinded to treatment allocation. (C) Schematic diagram of the animal experiment process. (D, F-H) Immunohistochemical staining and quantitative analysis show the expression changes of ADAMTS5, Aggrecan, and SOX9, scale bar: 200 μm. (E, I, J) Immunofluorescence staining and quantitative analysis show the expression changes of Collagen II and MMP13, scale bar: 200 μm. (K) Micro-CT three-dimensional reconstruction and cross-sectional images show the changes in joint structure, scale bar: 1 mm. (L, M) Quantitative analysis of trabecular thickness (Tb.Th) and bone volume/total volume (BV/TV) in the medial tibial subchondral bone. Only the DMM + GM@KP-ABs + US group received ultrasound stimulation at 1 MHz and 20 mW cm⁻² for 5 min every 2 days. Error bars show mean ± SD. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 6 denotes six animals per group.

Beyond the safety of ultrasound exposure, the long-term fate of KNN@PDA is also important for clinical translation. Although KNN is a lead-free piezoelectric ceramic with reported biocompatibility, it undergoes slow degradation in physiological fluids and may release alkali ions while leaving Nb-enriched residues [62]. PDA coating can reduce acute surface-associated toxicity [63] but may further delay particle degradation because of its prolonged in vivo stability [64]. Short-term intra-articular studies have reported no evident local or systemic toxicity from sodium potassium niobate nanoparticles [65]; nevertheless, the long-term biodistribution, chemical speciation, and clearance of KNN@PDA remain unresolved. Time-resolved elemental tracing and chronic toxicity studies are therefore required to establish its long-term safety before clinical translation. H&E and Safranin O/Fast Green staining were first performed to assess cartilage structure and proteoglycan preservation [66]. In the DMM group, articular cartilage exhibited typical degenerative features, including surface erosion, cartilage thinning, disorganized cartilage architecture, and markedly reduced Safranin O staining intensity (Fig. 8A). GM@KP treatment showed limited protection, whereas the GM@KP-ABs group partially preserved cartilage morphology and proteoglycan distribution. Notably, GM@KP-ABs + US treatment markedly improved cartilage structural integrity and partially recovered Safranin O staining, suggesting effective protection of cartilage matrix in vivo. Consistently, OARSI scoring showed that the GM@KP-ABs + US group significantly reduced cartilage degeneration compared with the DMM group and other treatment groups (Fig. 8B), indicating that the ultrasound-responsive composite hydrogel effectively attenuated DMM-induced cartilage degeneration.

Because proteoglycan loss is a hallmark of OA cartilage degeneration, we further examined ADAMTS5, a key aggrecanase responsible for aggrecan degradation during OA progression. Immunohistochemical staining showed that ADAMTS5 was strongly upregulated in the DMM group, while Aggrecan and SOX9 expression was substantially reduced, reflecting cartilage matrix degradation and impaired chondrocyte phenotype maintenance (Fig. 8D, F-H). GM@KP-ABs + US treatment significantly decreased ADAMTS5-positive staining and partially restored Aggrecan and SOX9 expression, indicating that cartilage catabolism is inhibited, while the anabolic function of chondrocytes (as evidenced by Aggrecan and SOX9) is preserved [67]. Immunofluorescence staining further confirmed this protective effect. Collagen II fluorescence was markedly reduced in the DMM group and was partially restored after GM@KP-ABs + US treatment, whereas MMP13 expression was strongly induced by DMM surgery and significantly decreased in the GM@KP-ABs + US group (Fig. 8E, I, J). These in vivo findings are consistent with the in vitro results, supporting the ability of ultrasound-responsive GM@KP-ABs to improve ECM-associated anabolic and catabolic markers under OA conditions.

Subchondral bone remodeling was further assessed by micro-CT. Compared with the Sham group, DMM surgery induced pronounced structural abnormalities in the knee joint, including irregular subchondral bone architecture and osteophyte-like bony remodeling (Fig. 8K). Quantitative analysis of the medial tibial subchondral bone showed that DMM significantly increased trabecular thickness (Tb.Th) and bone volume/total volume (BV/TV), consistent with pathological subchondral bone sclerosis during OA progression [68]. The GM@KP group and the GM@KP-ABs group partially alleviated these abnormalities, whereas GM@KP-ABs + US treatment more effectively reduced the abnormally elevated Tb.Th and BV/TV toward the Sham level, indicating attenuation of OA-associated subchondral bone remodeling (Fig. 8L and M). In addition, BMD analysis revealed that GM@KP-ABs + US treatment significantly reduced the elevated subchondral BMD compared with GM@KP-ABs alone, restoring the BMD level closer to that of the Sham group (Fig. S30). Because the present in vivo design does not separate ultrasound exposure, KNN@PDA-associated electromechanical activation, ultrasound-enhanced UA-AB release, and UA-AB bioactivity, the greater BMD reduction in the combination group should be interpreted as a combined treatment-associated improvement rather than a piezoelectric-specific effect. Together, these findings demonstrate that GM@KP-ABs + US treatment effectively attenuated pathological subchondral bone remodeling in OA.

The IL-1β-stimulated chondrocyte and RAW264.7 macrophage models provide controlled systems for mechanistic screening but do not reproduce the cellular heterogeneity and cross-tissue interactions of human OA. Similarly, although the DMM mouse model provides in vivo proof of concept, it primarily represents mechanically induced post-traumatic OA and does not fully reflect age-related or metabolic disease phenotypes. The absence of primary human OA cells or cartilage explants, together with species-related differences in joint size, cartilage thickness, and biomechanics, further limits the direct extrapolation of ultrasound delivery and therapeutic efficacy to humans. Future studies should therefore validate this strategy using primary human chondrocytes, synovial macrophages and fibroblasts, cartilage or osteochondral explants, aged and metabolic OA models, and large-animal models to establish its generalizability and translational potential.

Integrating the in vitro and in vivo findings, GM@KP-ABs + US treatment appears to exert therapeutic effects through coordinated biological and physical regulation. UA-ABs provide bioactive vesicular cues associated with mitochondrial quality control, while ultrasound activation of the KNN@PDA-containing hydrogel was associated with further improvement in mitochondrial functional indices. These complementary components may contribute to changes in the local immune microenvironment toward a more reparative state. At the tissue level, these effects translate into reduced cartilage degeneration, improved ECM-associated markers and anabolic-catabolic balance, and improved subchondral bone architecture. Collectively, the in vivo findings support GM@KP-ABs as a promising ultrasound-responsive therapeutic platform for attenuating OA progression. However, its clinical translation will require careful optimization of the ultrasound parameters according to patient- and joint-specific characteristics, as variations in tissue thickness, acoustic attenuation, joint geometry, and hydrogel distribution may affect the amount and spatial distribution of acoustic energy delivered to the implanted material. Therefore, validation in large-animal models, together with the establishment of standardized protocols for acoustic coupling, dosimetry, and image-guided treatment, will be essential before clinical application.

3. Conclusion

In this study, an injectable ultrasound-responsive hydrogel, GM@KP-ABs, was developed by incorporating KNN@PDA nanoparticles and UA-engineered apoptotic bodies into a GelMA matrix. GM@KP-ABs exhibited ultrasound-responsive electromechanical activity, ultrasound-enhanced release of UA-AB-associated components, and prolonged intra-articular retention. In IL-1β-stimulated chondrocytes, GM@KP-ABs combined with ultrasound improved mitochondrial function, increased PINK1/Parkin-associated mitophagic activity, suppressed cGAS-STING-NF-κB signaling, and reduced ECM catabolism. This treatment also induced an M2-like macrophage phenotype in vitro and increased F4/80⁺CD206⁺ macrophages in synovial tissue. In a DMM-induced mouse model of OA, GM@KP-ABs combined with ultrasound reduced cartilage degeneration and abnormal subchondral bone remodeling. These findings indicate that GM@KP-ABs regulates mitochondrial quality control and inflammation through ultrasound-responsive electromechanical and vesicle-associated effects and may represent a potential therapeutic approach for OA.

4. Experimental section

The detailed experiments and characterizations are listed in the Supplementary Information.

CRediT authorship contribution statement

Zhe Zhang: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Xiangheng Guan: Writing – original draft, Validation, Investigation, Formal analysis, Data curation. Yunhao Zhai: Visualization, Validation, Investigation, Formal analysis. Hongxin Lin: Visualization, Software, Formal analysis, Data curation. Anna Ying: Validation, Methodology, Investigation. Jiahao Lin: Methodology, Investigation, Data curation. Chenggong Zhu: Validation, Investigation, Formal analysis. Binquan Huang: Validation, Software, Formal analysis. Nuo Chen: Visualization, Investigation, Data curation. Kaihang Zhang: Writing – review & editing, Supervision, Resources, Funding acquisition, Conceptualization. Xuanyong Liu: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition. Jiahao Ying: Writing – review & editing, Supervision, Resources, Funding acquisition, Conceptualization.

Data availability

Data will be made available on request.

Ethics approval and consent to participate

All animal experiments were conducted in compliance with institutional guidelines for the care and use of laboratory animals and were approved by the Animal Welfare and Use Committee of Wenzhou Medical University, approval No. xmsq2025-0589.

Declaration of competing interest

Xuanyong Liu is an editorial board member for Bioactive Materials and was not involved in the editorial review or the decision to publish this article. All authors declare that there are no competing interests.

Appendix. ASupplementary data

The following is the Supplementary data to this article: