A honeysuckle-decorated nanoimmunomodulator hijacks Macrophage-BMSC crosstalk to promote arthritic regeneration Authors: Yuan Ma, Shan He, Hanzhong Liu, Jinye Zhang, Wenguo Huang, Yuan Gu, Naiqian Cui, Jikun Qian, Mincheng Lu, Daorong Xu, Zhiqing Li, Bin Yu Published: 2026 ABSTRACT AbstractOsteoarthritis (OA) is a chronic inflammatory joint disorder characterized by persistent synovitis and progressive cartilage degradation, driven primarily by dysregulated macrophage polarization and impaired chondrogenic differentiation of bone marrow mesenchymal stem cells (BMSCs). In this study, we innovatively developed a honeysuckle flavonoid-modified ruthenium nanocluster (RuNC@HS) nanozyme that exhibits dual superoxide dismutase (SOD) and catalase (CAT)-mimicking activities, enabling efficient reactive oxygen species (ROS) scavenging and immunomodulation. In vitro and in vivo experiments demonstrated that RuNC@HS inhibits NLRP3/caspase-1-mediated pyroptosis in BMSCs and blocks the subsequent HMGB1-induced activation of the TLR4/STAT3 pathway in macrophages, thereby disrupting a maladaptive pro-inflammatory feedback loop. These mechanisms significantly enhanced BMSC migration, chondrogenic differentiation, and M2 macrophage polarization. In a murine OA model induced by anterior cruciate ligament transection (ACLT), intra-articular administration of RuNC@HS markedly reduced osteophyte formation, restored subchondral bone architecture, improved gait function, and promoted cartilage matrix synthesis, while downregulating catabolic enzyme expression. Metabolomic profiling further revealed that RuNC@HS uniquely modulates anti-inflammatory lipid metabolism pathways, particularly α-linolenic acid metabolism. For the first time, this study highlights the dual innovative aspects of integrating traditional herbal medicine with precise nanozyme design and deeply deciphering BMSC-macrophage crosstalk as an immunometabolic regulatory strategy, offering a novel and synergistic therapeutic approach for OA.Keywords: Osteoarthritis, Ruthenium nanoclusters, Macrophage polarization, BMSC-Macrophage crosstalk, HMGB1/TLR4/STAT3 axis Introduction Osteoarthritis (OA) is a debilitating chronic inflammatory joint disease affecting over 500 million people globally, posing a substantial socioeconomic burden that accounts for 1.0–2.5% of the GDP in developed countries [1,2]. Its pathogenesis is characterized by persistent inflammation and progressive cartilage degradation [3], driven largely by immune dysregulation in which macrophages play a central role [[4], [5], [6]]. Although cartilage serves essential mechanical functions, its poor self-repair capacity is further compromised by inflammatory microenvironments that disrupt chondrocyte activity and impede regeneration [7,8]. In this context, bone marrow mesenchymal stem cells (BMSCs) have emerged as promising candidates due to their dual chondrogenic and immunomodulatory capacities—particularly their ability to mitigate inflammatory responses and promote cartilage repair [[9], [10], [11], [12], [13], [14]]. Synovial inflammation accelerates OA progression by activating macrophages and inducing chondrocyte apoptosis [15], with macrophage polarization (especially the M1/M2 balance) being closely correlated with disease severity [[16], [17], [18]]. While pro-inflammatory M1 macrophages exacerbate tissue damage through excessive ROS and cytokine production [[19], [20], [21], [22]], anti-inflammatory M2 macrophages promote cartilage anabolism and suppress inflammation [[23], [24], [25]]. Consequently, targeting macrophage reprogramming and leveraging BMSC-macrophage crosstalk have become compelling therapeutic strategies for modulating immune responses and restoring joint homeostasis in OA [26]. Nanozymes, particularly those mimicking antioxidant enzymes such as SOD, CAT, and GSHOx, have shown great potential in treating osteoarthritis (OA) through scavenging reactive oxygen species (ROS) and modulating macrophage polarization [[27], [28], [29], [30]]. Among them, ruthenium-based nanoclusters (Ru NCs), which consist of a few to hundreds of atoms with precise structural features (1–2 nm), exhibit remarkable catalytic activity, multi-enzyme mimicry, and favorable biocompatibility, rendering them promising anti-inflammatory nanomedicines [[31], [32], [33], [34], [35], [36], [37], [38], [39]]. Despite their efficacy in various inflammatory diseases, the application of Ru NCs in OA remains underexplored, and their performance is still limited by insufficient targeting ability, moderate anti-inflammatory efficacy, and lack of immunomodulatory synergy. To address these limitations, we propose a surface modification strategy using Lonicera japonica flavonoids (honeysuckle extract), a natural anti-inflammatory and antioxidant compound, to enhance the synovial targeting and macrophage-polarizing activity of Ru NCs. This study aims to develop a bio-enhanced Ru NC-based nanozyme and systematically evaluate its therapeutic effects and mechanism in OA management. To address these limitations, we employed a surface modification strategy using bioactive flavonoids derived from Lonicera japonica (honeysuckle). This selection was based on several rational considerations beyond its well-documented anti-inflammatory and antioxidant properties [40,41]. Firstly, honeysuckle flavonoids—primarily luteoloside and chlorogenic acid—possess specific hydroxyl and carbonyl motifs that facilitate stable coordination with ruthenium atoms [42], forming a uniform hybrid nanostructure (RuNC@HS) with enhanced catalytic stability. Secondly, compared to other prominent anti-OA TCM components such as curcumin (poor water solubility and stability) and berberine (potential cytotoxicity at higher doses), honeysuckle flavonoids offer superior aqueous compatibility and a established safety profile, making them more suitable for nanozyme functionalization [43]. This dual-pathway regulatory potential is less pronounced in icariin or other single-component TCM extracts, positioning honeysuckle flavonoids as an ideal candidate to confer multi-target immunomodulatory synergy to RuNCs. In this study, we developed a unique ruthenium-based nanozyme—honeysuckle flavonoid-modified ruthenium nanoclusters (RuNC@HS)—that exhibits dual enzyme-mimetic activities, mimicking both superoxide dismutase (SOD) and catalase (CAT), enabling efficient reactive oxygen species (ROS) scavenging and immunomodulation in osteoarthritis (OA). Although nanomaterials have shown broad application potential in inflammatory diseases, the therapeutic promise of integrated traditional Chinese and Western catalytic nanoclusters in OA remains underexplored, particularly regarding their ability to modulate the crosstalk between bone marrow mesenchymal stem cells (BMSCs) and macrophages within the synovial inflammatory microenvironment. Functionalization of RuNCs with honeysuckle flavonoids (RuNC@HS) significantly enhanced their anti-inflammatory efficacy, improved synovial bioavailability, and augmented macrophage reprogramming capability. Through integrated transcriptomic, metabolomic, and functional analyses, we systematically evaluated the capacity of RuNC@HS to reshape the inflammatory microenvironment and promote cartilage regeneration. We further demonstrated that RuNC@HS alleviates OA progression by suppressing NLRP3/caspase-1-mediated pyroptosis in BMSCs and inhibiting HMGB1/TLR4/STAT3 pathway-driven pro-inflammatory crosstalk in macrophages, ultimately facilitating OA recovery. Results and discussion Synthesis of RuNC@HS Scme 1.a illustrates the synthetic scheme for Ru nanoclusters (RuNCs) and RuHC@HS. Briefly, mesoporous silica nanoparticles (MSNs) doped with varying Ru concentrations were coated with a polymeric layer and pyrolyzed at 900 °C, followed by hydrogen peroxide leaching to yield RuNCs. The dispersed RuNC were stirred to get RuNC@HS. For comparison, Ru-free nitrogen-doped porous carbon (Np) was prepared under identical conditions. The successful synthesis and structural characterization of RuNC@HS were systematically investigated. The dual-phase architecture was confirmed by transmission electron microscopy (TEM, Fig. 1A), which revealed uniformly distributed nanoparticles without significant aggregation. High-resolution TEM (HRTEM, Fig. 1B) further evidenced the precise control over ruthenium nanocluster formation achieved by the synthetic strategy. Aberration-corrected TEM imaging under higher magnification clearly resolved lattice fringes of the ruthenium nanoclusters, confirming the presence of well-defined nanocrystals (NCs) (Fig. 1C). The crystalline nature of the catalysts was examined by X-ray diffraction (XRD, Fig. 1D). RuNCs exhibited a weak broad peak centered at approximately 40°, corresponding to the Ru (101) plane, consistent with the cluster-type lattice structure observed in Fig. 1C. Elemental mapping (Fig. 1E) demonstrated a homogeneous spatial distribution of Ru, Si, N, and C, confirming the effective anchoring of Ru sites to the carbon matrix derived from pyrolyzed mesoporous silica nanoparticles (MSNs). Sub-nanometric Ru clusters (highlighted with yellow lattices) were dominant in RuNCs. X-ray photoelectron spectroscopy (XPS, Fig. 1F and G) provided further electronic-state evidence: the N 1s spectra indicated that graphitic N (located at 401.2 eV) served as the primary anchoring site for RuNCs, facilitating cluster formation. In the XPS analysis, the significantly higher proportion of C–C/C–H bonds (57.81%) compared to C–O/C–N (29.07%) and C=C (15.12%) suggests a highly graphitized carbon framework with enhanced electronic conductivity and chemical stability. This carbon environment facilitates efficient electron transfer during catalytic processes, thereby contributing to the superior catalytic performance of the nanoclusters. The dominant presence of sp3 and aliphatic carbon moieties may also promote better dispersion and anchoring of active metal sites, further optimizing the catalytic activity. Specifically, the assembly sites within the ruthenium nanoclusters promote multi-electron transfer processes, which is a key factor underlying the high catalytic activity of RuNC@HS observed in subsequent studies. Characterization of RuNC@HS. (A) Transmission Electron Microscopy (TEM) images and (B) high resolution transmission electron microscopy (HRTEM) of RuNC@HS. (C) Spherical Aberration Corrected Transmission Electron Microscope (AC-TEM) image of RuNC@HS, with isolated Ru cluster-like lattices highlighted by yellow circles. (D) X-ray diffraction (XRD) patterns of honeysuckle (HS) and RuNC@HS. (E) Corresponding High-angle annular dark-field scanning TEM (HAADF-STEM) and TEM-energy-dispersive X-ray spectroscopy elemental mapping (TEM-EDS) images of RuNC@HS. X-ray photoelectron spectroscopy (XPS) survey spectra: (F) High-resolution XPS spectra of N 1s. (G) High-resolution XPS spectra of C 1s. (H) Ultraviolet (UV) absorption spectrum of HS and RuNC@HS. (I) Nanoparticle diameter of HS and RuNC@HS via DLS. (J) Zeta potential. (K) Ultraviolet detection of HS content at different time points and in simulated microenvironments. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Subsequent UV absorption spectroscopy revealed a characteristic peak of honeysuckle (HS) at 266 nm in RuNC@HS, confirming the successful loading of HS (Fig. 1H). Quantitative analysis using the aluminum chloride colorimetric assay determined the honeysuckle flavonoid content loaded onto the nanoclusters to be 35.72 wt% (Fig. S1, supplementary information). Dynamic light scattering (DLS) measurements indicated that the hydrodynamic sizes of RuNC@HS and RuNC were approximately 153 nm and 166 nm, respectively (Fig. 1I). Zeta potential results further demonstrated that HS, RuNC, and RuNC@HS all exhibited relatively stable colloidal states (Fig. 1J). In healthy joints, synovial fluid exhibits a weakly alkaline pH ranging from 7.3 to 7.7, similar to blood, creating an ideal environment for cellular functions. However, in osteoarthritis (OA), the joint pH drops significantly to 6.0–7.0 or even lower, resulting in an acidic microenvironment with increased protein and mucus content. Therefore, when designing nanomedicines for OA treatment, the impact of pH reduction on drug release should be considered. To evaluate pH-dependent release behavior, the HS accumulation release profiles of RuNC@HS under different pH conditions were monitored over time using UV spectroscopy. The results showed that sustained HS release in the simulated arthritic microenvironment was significantly higher than in simulated normal joint lubrication fluid (Fig. 1K). Furthermore, it was confirmed that RuNC@HS maintains dimensional and aggregation stability in both serum and physiologically simulated environments (Fig. S2, supplementary information). Structure of RuNCs The local coordination environment and electronic structure of ruthenium species in the RuNCs were systematically investigated using X-ray absorption spectroscopy (XAS). As shown in Fig. 2A and Table S3 (Supporting Information), the Fourier-transformed EXAFS spectrum of RuNCs exhibits a distinct Ru–Ru coordination peak (∼2.5 Å), indicating the presence of metallic Ru aggregation. EXAFS fitting analysis further revealed a slightly shorter Ru–Ru bond distance in RuNCs (2.68 Å, Fig. 2D) compared to bulk Ru foil (2.70 Å), suggesting the presence of strain effects within the sub-nanometer clusters. Structure of RuNC. (A) k3-weighted Fourier transformed EXAFS function spectra of different samples. (B) Normalized XANES spectra and reference samples (Ru Foil, RuO2). (D-F) EXAFS fitting in R space. (G-I) Wavelet transformation of Ru Foil, RuO2 and RuNC. X-ray absorption near-edge structure (XANES) spectroscopy was employed to probe the electronic state of Ru species (Fig. 2B). The absorption edge position of RuNCs was found to be closer to that of metallic Ru foil, yet with a attenuated white-line intensity, reflecting their metallic nature and modified d-band electronic structure induced by quantum confinement effects. Quantitative EXAFS fitting (Fig. 2C–F) indicated that the Ru atoms in RuNCs have a coordination number (CN) of approximately 4 for Ru–N bonds, which is significantly lower than that of bulk Ru (CN = 12), consistent with their sub-nanometric size. Wavelet transform (WT) analysis (Fig. 2G–I) provided further spatial-frequency resolution: RuNCs showed high-intensity WT contours around 9 Å−1 corresponding to Ru–Ru contributions, which are distinct from the extended atomic ordering observed in Ru foil. Additionally, the WT maxima of RuNCs were shifted relative to those of RuO2, ruling out significant oxide formation and supporting the dominance of nitrogen-coordinated Ru centers within the clusters. Enzyme-mimetic ROS scavenging mechanism and biocompatibility of RuNC@HS The reactive oxygen species (ROS)-scavenging capabilities of RuNC and RuNC@HS were systematically evaluated through superoxide dismutase (SOD)-like and catalase (CAT)-like catalytic activities. Although both nanomaterials decompose H2O2 via electron transfer, their atomic structures govern distinct mechanistic pathways (Fig. 3A). TEM imaging further revealed structural deformation in RuNC@HS under H2O2 exposure, indicative of surface oxidation during catalytic turnover (Fig. 3B). RuNC@HS demonstrated remarkable SOD-like activity, which is displayed higher activity than RuNC, facilitated by metallic Ru–Ru sites that promote H2O2 dissociation (2H2O2 → 2H2O + O2) (Fig. 3C–D). Meanwhile, compared with RuNCs, RuNC@HS exhibited enhanced CAT-like activity. This was specifically demonstrated by increased dissolved oxygen levels under coexistence conditions (Fig. 3E), and ELISA analysis revealed higher CAT enzyme activity measurements (Fig. 3F). These results may be attributed to the inherent antioxidant properties of honeysuckle (HS) components in the loaded formulation. Catalytic performance and biosafety of RuNC@HS. (A) Schematic illustration of reactive oxygen species (ROS) scavenging. (B) Transmission electron microscopy (TEM) images showing structural deformation of nanoparticles in the presence of H2O2. (C, D) SOD-like and (F) CAT-like catalytic activities during H2O2 decomposition, measured by UV-absorption spectroscopy using ELISA kits. (E) Dissolved oxygen (DO) level changes in water under catalase (CAT)-like activity. (G) Internalization of Rho-labeled RuNC and RuNC@HS by endothelial cells observed via Confocal fluorescence images of Rho-labeled nanoparticles internalized by BMSCs under co-culture conditions; (H) corresponding quantitative data. (I) Hemolysis assay. (J) Cell viability of various repair cells co-cultured with RuNCs, assessed by CCK-8 assay. (K) ROS levels in different groups measured using DCFH-DA probes, and (L) statistical analysis. Critically, RuNC@HS showed excellent biocompatibility. Confocal microscopy confirmed the efficient and time-dependent internalization of Rho-labeled nanoparticles into bone marrow stromal cells (BMSCs), with even intracellular distribution (Fig. 3G and H). A concentration-dependent assessment of biocompatibility confirmed that RuNC@HS exhibited no significant hemolytic activity (Fig. 3I) and maintained cell viability above 90% in both BMSCs and macrophages across a wide concentration range, up to a maximum of 100 mg/mL (Fig. 3J). Low cytotoxicity and minimal metal ion leaching are associated with the dispersion of atomic metal clusters that slow down the associated cellular stress. Additionally, HS as a lipophilic envelope enhances endocytosis and cellular compatibility. In osteoarthritis animal models, we administered RuNC@HS nanoparticles via a single injection once a week. By the 8th week, blood, joint tissue, and samples from major organs (liver, spleen, kidneys, heart, and lungs) were collected. Quantitative analysis of ruthenium (Ru) in tissues and blood showed negligible Ru residues. Pathological sections further demonstrated no nanoparticle accumulation in vital organs or joint tissues, confirming the material's long-term biosafety (Fig. S3, supplementary information). To assess functionality in a biologically relevant context, intracellular ROS levels were probed using DCFH-DA. In inflammatory conditions, RuNC@HS reduced ROS in BMSCs by 52%, outperforming RuNC, which achieved a 37% reduction (Fig. 3K and L). This enhanced scavenging efficiency of RuNCs is attributed to their more balanced SOD-CAT synergistic activity, which effectively neutralizes early-phase O2•- radicals through SOD-like activity and clears H2O2 accumulation via dominant CAT-like activity in later stages. These results underscore the role of structural configuration in determining enzymatic performance and establish RuNC@HS as a highly efficient and biosafe nanozyme capable of operating in oxidative microenvironments. Attenuation of cartilage degradation and promotion of chondrogenesis in an inflammatory microenvironment Cartilage degeneration is associated with a variety of pathogenic factors, including inflammation, aging, oxidative stress, and mechanical trauma [44]. Due to the avascular nature of cartilage and limited proliferative capacity of chondrocytes, intrinsic regenerative ability is severely restricted. Bone marrow-derived mesenchymal stem cells (BMSCs), recognized for their chondrogenic potential and immunomodulatory properties, have been widely employed to promote cartilage repair and modulate inflammatory microenvironments [45]. Although BMSCs are scarcely present within the articular cavity under physiological conditions, they become activated and contribute to cartilage regeneration following injury [46]. In this study, we first co-cultured BMSCs with various nanozymes under inflammatory stimulation. Wound healing assays demonstrated that lipopolysaccharide (LPS) significantly impeded cell migration, whereas treatment with RuNC@HS partially restored migratory capacity (Fig. 4A and B). The migration rate in the LPS + RuNC@HS group exceeded those in the LPS, LPS + Np, and LPS + RuNC groups, approaching the level observed in the control group. Furthermore, both RuNC@HS and RuNC treatments enhanced BMSC proliferation under inflammatory conditions (Fig. 4C and D), suggesting a protective role against inflammation-induced impairment. RuNC@HS regulates BMSCs proliferation and chondrogenic differentiation. G1: BMSCs did not receive any treatment; G2: BMSCs were induced by 5 μg/mL LPS to stimulate OA; G3, G4, G5: the established LPS-induced BMSCs were treated with Np, RuNC and RuNC@HS, respectively. (A, B) Representative images (A) and wound healing rates (healed area/scratch area %) (B) in the scratch assay showing BMSCs migration. (C, D) F-actin staining in BMSCs (C) and quantitative fluorescence intensity (D). Scale bar: 100 μm. (E) Safranin O‐Fast Green staining and Alcian blue staining, and (F, G) their quantitative analysis. (H) CLSM images of COL2A1/MMP13 immunofluorescence and (I, K) their fluorescence intensity quantification. (J) RT‐PCR results showing the relative mRNA expression levels of Col II. Data are expressed as mean ± SD (n ≥ 3). One-way ANOVA was used to analyze the statistical differences between the 2 groups, and ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ns = no significance. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) To evaluate chondrogenic matrix synthesis, we performed Alcian blue and Safranin O–Fast Green staining. As shown in Fig. 4E, the RuNC@HS group exhibited abundant and densely stained proteoglycan deposits, similar to the control group. In contrast, the LPS and LPS + Np groups showed only sparse matrix staining, while the LPS + RuNC group displayed moderate improvement. Quantitative analysis of Alcian blue staining confirmed that RuNC@HS treatment significantly increased extracellular matrix content compared to all other LPS-induced groups, though it remained lower than that in the non-inflammatory control (Fig. 4F). Similarly, Safranin O staining intensity was markedly elevated in the RuNC@HS group (Fig. 4E–G), indicating enhanced proteoglycan accumulation and chondrogenic differentiation. Immunofluorescence analysis further revealed that RuNC@HS treatment markedly upregulated the expression of COL2A1, a key cartilage collagen component (Fig. 4H–J), and suppressed the catabolic enzyme MMP13 under inflammatory conditions (Fig. 4K). These expression levels were comparable to those in the control group, suggesting that RuNC@HS effectively promotes anabolic activity and inhibits matrix degradation in BMSCs. In summary, these results demonstrate that RuNC@HS nanozymes mitigate LPS-induced inflammatory impairment by enhancing BMSC migration, proliferation, and chondrogenic matrix deposition, highlighting their potential as a therapeutic strategy for cartilage regeneration. RuNC@HS-treated BMSC secretome attenuates macrophage inflammation via the HMGB1/TLR4/STAT3 pathway The immunomodulatory crosstalk between bone marrow mesenchymal stem cells (BMSCs) and macrophages plays a pivotal role in the progression of arthritis [47,48]. Within the inflammatory microenvironment, BMSCs secrete cytokines that promote macrophage polarization from a pro-inflammatory M1 toward an anti-inflammatory M2 phenotype [[49], [50], [51]], thereby alleviating inflammation and facilitating tissue repair. Targeting this interplay represents a promising therapeutic strategy for arthritis. Building on previous evidence that RuNC@HS enhances BMSC proliferation, osteogenic capacity, and collagen repair, we further explored its anti-apoptotic and anti-inflammatory mechanisms under inflammatory challenge. Although RuNC@HS outperformed RuNC in promoting chondrogenesis and proliferation under inflammatory conditions, its molecular mechanisms remained unclear. Transcriptomic profiling of inflamed BMSCs treated with RuNC@HS versus RuNC identified 1160 differentially expressed genes (DEGs), including 1081 upregulated and 79 downregulated (P < 0.05, |log2FC| > 1.5; Fig. 5Ai). KEGG analysis indicated significant enrichment in pathways related to reactive oxygen species, arthritis, NF-κB signaling, and apoptosis (Fig. 5Aii). GO and Reactome analyses further revealed associations with cellular metabolism, proliferation, immune regulation, and antioxidant defense (Fig. 5Aiii). Notably, RuNC-treated cells exhibited upregulation of NF-κB, JAK-STAT, TNF, and MAPK pathways—key inflammatory cascades targeted in antioxidant therapy for arthritis. RuNC@HS regulates macrophage polarization and modulates macrophage inflammatory responses via the HMGB1/TLR4/STAT3 pathway. G1: RAW264.7 exposed to BMSC-conditioned medium without any treatment; G2: RAW264.7 were treated with LPS-induced BMSC-conditioned medium; G3, G4, G5: RAW264.7 cells treated with conditioned medium from BMSCs sequentially stimulated by LPS and Np/RuNC/RuNC@HS nanozymes, respectively. (A, i) Volcano plot of the distribution of DEGs between Control and LPS + RuNC@HS group. (A, ii) Bar diagram of partial biology progress KEGG term clusters of DEGs. (A, iii) Bubble diagram of partial significantly enriched pathways for DEGs. (B–E) Representative immunoblot analysis and quantification of the NLRP3 signaling pathway. (F) Bar graph showing the relative expression level of HMGB1 in Control and LPS + RuNC@HS groups. (G) Representative immunoblot analysis and quantification of HMGB1 protein expression. (H) Schematic diagram of macrophage modulation by different-treated BMSC-conditioned medium. (I, J) Representative western blot analysis of HMGB1/TLR4/STAT3 pathway. (K) CLSM images of Arg1/iNOS immunofluorescence with bright-field microscopy and (L) fluorescence intensity quantification. (M − O) RT‐PCR results showing the relative mRNA expression levels of pro-inflammatory cytokines. Data are expressed as mean ± SD (n ≥ 3). One-way ANOVA was used to analyze the statistical differences between the 2 groups, and ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ns = no significance. Given these findings, we investigated whether RuNC@HS modulates the NLRP3/Caspase-1 pathway, a critical mediator of pyroptosis and sterile inflammation. NLRP3 activation promotes Caspase-1 cleavage [52], leading to maturation of IL-1β and IL-18 and amplification of local inflammation [53,54]. Our results confirmed that LPS robustly induced NLRP3 and Caspase-1 activation in BMSCs (Fig. 5B–E). Since Caspase-1-dependent pyroptosis triggers release of HMGB1 [55]—a damage-associated molecular pattern that exacerbates inflammation [56]—we evaluated whether RuNC@HS influences this axis. Transcriptomic and Western blot analyses consistently demonstrated that RuNC@HS treatment significantly suppressed NLRP3 activation, Caspase-1 cleavage, and HMGB1 expression (Fig. 5F and G and S4A). These results indicate that RuNC@HS attenuates pyroptosis-driven HMGB1 release by inhibiting the NLRP3/Caspase-1 pathway, thereby disrupting a key inflammatory circuit and supporting a regenerative microenvironment. It is now generally accepted that paracrine secretion is the main cause of mutual crosstalk between BMSCs and macrophages [57,58]. However, the mechanisms of RuNC@HS nanozymes impacting the mutual crosstalk in living organisms to improve homeostasis for OA still remains obscure. We further detected the interaction effect of RuNC@HS on BMSCs-macrophage in vitro. BMSCs were incubated in PBS or LPS for 2 days after pre-cultured with Np, RuNC, RuNC@HS nanomyzes for 12 h. Macrophages were seeded into the cell dishes and incubated at 37 °C for 48 h with supernatants of BMSCs treated under various co-coculture conditions as described in scheme (Fig. 5H). Given the established role of HMGB1 as a critical mediator of inflammation and its ability to activate TLR4 and downstream STAT3 signaling—leading to amplified cytokine production [59,60]—we asked whether RuNC@HS influences this pathway. Western blot analysis demonstrated that macrophages treated with supernatant from RuNC@HS-treated BMSCs exhibited significantly lower protein levels of HMGB1, TLR4, and phosphorylated STAT3 compared to those treated with supernatants from the Np- or RuNC-groups (Fig. 5I and J; Fig. S4B–E, Supporting Information). Furthermore, immunofluorescence staining revealed that supernatant from LPS-stimulated BMSCs strongly induced iNOS expression in macrophages—a marker of the pro-inflammatory M1 phenotype. In contrast, supernatants from both RuNC- and RuNC@HS-treated BMSCs suppressed iNOS expression. Importantly, the RuNC@HS group exhibited the most pronounced induction of Arg1, a characteristic M2 marker, indicating a robust shift toward anti-inflammatory macrophage polarization (Fig. 5K and L). Consistent with this suppression of the HMGB1/TLR4/STAT3 axis, we observed a marked reduction in the secretion of pro-inflammatory cytokines IL-1β and IL-6, alongside a significant increase in the anti-inflammatory cytokine IL-10 (Fig. 5M−O). To obtain more robust evidence, we isolated and cultured mouse bone marrow mesenchymal stem cells (BMSCs) for subsequent experiments. Intracellular superoxide anion levels were assessed in situ using Dihydroethidium (DHE) fluorescence staining. As shown in Fig. S5A–B, compared to the control group, the LPS-treated group exhibited a significantly enhanced red fluorescence signal, whereas treatment with RuNC@HS markedly reduced intracellular superoxide levels. These results indicate that RuNC@HS effectively catalyzes peroxide decomposition and ameliorates cellular oxidative stress. We further investigated whether RuNC@HS exerts its therapeutic effects by modulating the NLRP3/Caspase-1 pathway—a key axis mediating pyroptosis and sterile inflammation. Results in Fig. S5 demonstrate that LPS significantly activated NLRP3 and Caspase-1 in BMSCs. In contrast, RuNC@HS treatment effectively suppressed NLRP3 activation, Caspase-1 cleavage, and downstream HMGB1 release. These findings confirm that RuNC@HS inhibits the NLRP3/Caspase-1-mediated pyroptosis pathway, thereby blocking HMGB1 release and disrupting a critical inflammatory circuit, which favors the establishment of a regenerative microenvironment. To establish the causal role of the HMGB1/TLR4/STAT3 signaling pathway in the therapeutic mechanism of RuNC@HS, we performed rescue experiments in a co-culture system of mouse BMSCs and macrophages. The results demonstrated that targeted inhibition at multiple nodes of this signaling cascade completely abolished the therapeutic benefits of RuNC@HS, indicating that its protective effects are closely associated with this pathway (Fig. S6, supplementary information). Specifically, neutralization of extracellular HMGB1 with a specific antibody significantly attenuated the inhibitory effect of RuNC@HS on the expression of key pro-inflammatory mediators such as NLRP3 and iNOS. Similarly, pharmacological inhibition of TLR4, the receptor for HMGB1, using TAK-242, largely reversed the therapeutic outcomes. Most conclusively, direct knockdown of STAT3 using siRNA completely abrogated the efficacy of RuNC@HS, as reflected by restored high levels of inflammatory cytokines and loss of anti-inflammatory macrophage polarization. Collectively, these loss-of-function experiments provide direct evidence that RuNC@HS alleviates osteoarthritis progression primarily via the HMGB1/TLR4/STAT3 signaling axis. Consistently, in downstream macrophages, RuNC@HS restored the upregulation of M2 macrophage-associated genes (Arg1), suppressed M1-associated gene expression (iNOS), and reversed the LPS-induced elevation of pro-inflammatory cytokines (IL-1β, IL-6) while promoting the anti-inflammatory cytokine IL-10 (Fig. S7, supplementary information). Collectively, these findings demonstrate that RuNC@HS preconditioning enhances the paracrine function of BMSCs, enabling their secretome to effectively suppress the HMGB1/TLR4/STAT3 pathway in macrophages, mitigate inflammatory responses, and promote a pro-reparative M2 phenotype—highlighting a novel nanozyme-mediated strategy to modulate cellular crosstalk in OA. In vivo therapeutic efficacy of RuNC@HS in attenuating osteoarthritis progression Building on its promising in vitro anti-inflammatory performance, RuNC@HS nanozymes were further assessed for inhibiting the progression of OA in vivo. As evidenced by the microcomputed tomography (micro‐CT) scans, the RuNC and RuNC@HS nanozymes treated groups exhibited reduced knee osteophyte formation compared to the positive control group, with the RuNC@HS nanozymes group showing significantly fewer osteophytes, comparable to sham controls. In contrast, prominent osteophytes were consistently observed in both positive control and Np nanozymes treated groups. Both the positive control and Np nanozymes treated groups exhibited a distinct high bone mass phenotype in subchondral bone, which was significantly ameliorated in RuNC and RuNC@HS nanozymes treated groups (Fig. 6Aii). Tibial subchondral bone volume (BV)/tissue volume (TV) ratio, trabecular pattern factor (Tb.Pf) and subchondral trabecular bone thickness (Tb.Th) increased dramatically in PBS and Np group. Because of the fusion of the trabeculae, subchondral trabecular number (Tb.N) was reduced in PBS and Np group (Fig. 6F–I). Triplanar CT visualization all consistently supported these findings (Fig. 6B). Therapeutic assessment of RuNC@HS for OA management. (A) Representative micro-CT images of knee joints and knee joint subchondral bones in sham group, ACLT group, ACLT + Np group, ACLT + RuNC group and ACLT + RuNC@HS group. (B) Triplanar view of micro-computed tomography scanning of joint subchondral bone. (C) Representative images of HE staining, Alcian blue staining and Safranin O‐fast green staining in each group. (D, E) Representative gait screen shot and gait intensity. (F-I) Quantitative analysis of Tb.N (trabecular number), Tb.Th (trabecular thickness), Tb.pf (trabecular pattern factor) and BV/TV (bone tissue relative to the total tissue volume) in each group. (J) OARSI score of each group. (K) The ratio of hyaline cartilage (HC) and calcified cartilage (CC) of articular cartilage. (L) Quantitative analysis of relative glycosaminoglycan (GAG) content. (M) Quantification of Rear Right limb average brake time. (N) Quantification of Rear Right limb average print area. (O) Quantification of Rear Right limb average stride length. (P) Quantification of Rear Right limb average stride time. (Q) Quantification of Rear Right limb average stance time. Data are expressed as mean ± SD (n ≥ 3). FR: front right limb. FL: front left limb. RR: rear right limb. RL: rear left limb. One-way ANOVA was used to analyze the statistical differences between the 2 groups, and ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ns = no significance. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) To further ascertain the potential of RuNC@HS in retarding the progression of OA, histological staining assays were performed on cartilage tissues, encompassing a range of techniques such as hematoxylin‐eosin (H&E) staining, Safranin O‐fast green staining and Alcian blue staining. As illustrated in Fig. 6C, the ACLT group exhibited characteristic cartilage features of OA, including surface discontinuity and erosion fissures. The group treated with Np showed no notable differences compared to the ACLT group. However, both RuNC and RuNC@HS groups demonstrated significant enhancements in morphological changes, matrix staining, and tidemark integrity. RuNC@HS was regarded as the most efficacious treatment option, exhibiting the most favorable OARSI score compared with the other treatments (Fig. 6J). In addition, the depth of the cartilage lesions was significantly diminished in the RuNC and RuNC@HS groups in comparison to the ACLT and Np groups, as evidenced by the HC/CC ratio and relative glycosaminoglycan (GAG) content of each group. Notably, the RuNC@HS group displayed the best results in terms of the HC/CC ratio and relative GAG among the various treatment groups, exhibiting a significant increase relative to that of the ACLT group (Fig. 6K and L). To further evaluate the therapeutic efficacy of RuNC@HS nanozymes in vivo, we established an ACLT model in the right hindlimb of mice and conducted gait analysis 8 weeks after intra-articular nanozymes administration. The gait index, encompassing gait screenshots and gait intensities, revealed that ACLT mice exhibited higher gait intensities compared with sham samples (Fig. 6D and E). To investigate the changes across all treatment groups, an independent analysis was conducted specifically on the rear right limb that underwent intra-articular injection with nanozymes. This analysis yielded parameters indicative of right hind limb function, including average print area, mean stride length, mean brake time, mean stride time, mean stance time (Fig. 6M–Q). Due to cartilage damage and joint pain, ACLT and Np group mice exhibited gait abnormalities marked by a decrease in print area, a shortening of stride length, an elongation of stride time, a prolongation of brake time, and an extension of stance time, indicating that reduction of walking speed. Conversely, RuNC@HS exhibited the most pronounced therapeutic improvements across measured parameters. These findings suggested that intra-articular injection of RuNC@HS not only did not impede knee motion but also significantly improved knee joint function and gait patterns in OA mice. RuNC@HS nanozymes modulate macrophage polarization and metabolic pathways to attenuate OA progression To evaluate the metabolic equilibrium within cartilage, we performed immunohistochemical and immunofluorescent staining. The RuNC@HS group exhibited the most favorable outcomes, characterized by a substantial decrease in MMP13 levels (Fig. 7A and B) and significant upregulation of collagen type II (Col II) expression (Fig. 7C and D) compared to other treatment groups. These results indicate that RuNC@HS nanozymes effectively mitigate cartilage degradation and delay OA progression. RuNC@HS alleviates effectively cartilage degeneration and attenuates OA progression in vivo. G1: Joint tissues isolated from Sham-treated mice treated with PBS; G2: Joint tissues isolated from ACLT-treated mice treated with PBS; G3, G4, G5: Joint tissues isolated from ACLT-treated mice treated with Np/RuNC/RuNC@HS nanozymes respectively. (A) Immunofluorescence staining to detect the expression of MMP13 in different groups, and (B) corresponding analysis of the percentage of MMP13‐positive expressing cells in cartilage layer. (C) Immunohistochemical staining for Col II for assessing the metabolic balance of cartilage, and (D) quantitative analysis. (E) Immunofluorescence staining to detect the expression of iNOS/CD206 (left panel) with enlarged area in the right box of the synovium tissue area (right panel). (F) Quantitative analysis of Fig. 7E. (G) RT‐PCR results showing the relative mRNA expression levels of macrophage polarization. (H-I) RT‐PCR results showing the relative mRNA expression levels of inflammatory cytokines. (J) PCA score plot of knee joint samples from Ru-SA and Ru-NC mice groups; (K) Violin plots showing the levels of antioxidant-related fatty acids in response to nanozyme treatments. (L) KEGG bubble plot of differentially metabolites. (M, N) Representative immunoblot analysis of the NLRP3 signaling pathway in joint tissues mentioned above. Data are expressed as mean ± SD (n ≥ 3). One-way ANOVA was used to analyze the statistical differences between the 2 groups, and ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ns = no significance. We next assessed macrophage infiltration in synovial tissues using immunofluorescence staining for iNOS (M1 marker) and CD206 (M2 marker) (Fig. 7E). Intense iNOS staining was observed in the ACLT group, while both RuNC and RuNC@HS treatments significantly reduced iNOS expression, with RuNC@HS showing the most pronounced reduction (Fig. 7F). Conversely, CD206 expression demonstrated an inverse pattern, with the RuNC@HS group exhibiting the strongest M2 macrophage polarization. These findings were further validated by qPCR analysis of joint tissues (Fig. 7G). Consistent with macrophage polarization shifts, pro-inflammatory cytokines (IL-6, TNF-α) were significantly downregulated, while anti-inflammatory cytokines (IL-4, IL-10) were upregulated in the RuNC@HS group (Fig. 7H and I). To gain further insight into the metabolic alterations induced by RuNC@HS, we conducted untargeted metabolomic analysis of knee joint tissues. Principal Component Analysis (PCA) revealed clear separation between the RuNC and RuNC@HS groups (Fig. 7J). Violin plot analysis identified significantly elevated levels of α-linolenic acid (ALA, ω-3 PUFA) and reduced levels of arachidonic acid (AA, ω-6 PUFA) in the RuNC@HS group (Fig. 7K). ALA is known to possess potent anti-inflammatory and antioxidant properties, inhibiting pro-inflammatory cytokine production and protecting against cellular damage, while AA-derived lipid mediators play complex roles in inflammation resolution [61]. ALA protects against the proliferation of pro-inflammatory cytokines, including nitric oxide synthase and cyclooxygenase-2 (COX-2) [62,63]. It has also a strong antioxidant potential and free radical scavenging activity to protect against cellular damage, apoptosis, and attenuates inflammatory response [64,65]. Omega 6 is an essential fatty acid that plays a vital role in many physiological functions. Lipid mediators originated by AA, are key factors of the resolution of inflammation, once they orchestrate the clearance of pro-inflammatory cells and signals promoting tissue restoration [66]. KEGG pathway enrichment analysis further revealed significant alterations in arachidonic acid metabolism, alpha-linolenic acid metabolism, and glycerophospholipid metabolism (Fig. 7L), all of which are critically involved in inflammatory regulation. Fatty acid desaturase 2 (FADS2) is an endoplasmic reticulum-localized rate-limiting enzyme that regulates the balance between ω-3 and ω-6 pathways by catalyzing double-bond insertion in fatty acid chains [67]. Its activity determines the arachidonic acid (AA) to EPA/DHA ratio—AA being a precursor to pro-inflammatory mediators, and EPA/DHA yielding anti-inflammatory and pro-resolving mediators [68]. Thus, enhancing FADS2 activity is a potential strategy to rectify ω-6/ω-3 imbalance and suppress inflammation at its origin. However, chronic inflammation often impairs FADS2 function, leading to blocked ω-3 metabolism and sustained pro-inflammatory status [69,70]. Notably, RuNC@HS treatment restored FADS2 expression and function, essential for ω-3 fatty acid conversion and elongation, while simultaneously suppressing the pro-inflammatory enzymes COX-2 and 5-LOX. These results indicate that RuNC@HS drives a favorable ω-3/ω-6 shift via a dual mechanism: promoting FADS2-mediated ω-3 synthesis while inhibiting the COX-2/5-LOX-mediated inflammatory pathways derived from AA (Fig. S8, Supporting Information). Within the AA cascade, COX-2 and 5-LOX are core inflammatory enzymes [71]. COX-2, induced in pathologies such as rheumatoid arthritis, generates prostaglandin E2 (PGE2), a potent pro-inflammatory lipid [72]. 5-LOX converts AA into leukotrienes including LTB4, which recruits and activates inflammatory cells. Together, they amplify and sustain inflammatory progression [73,74]. Finally, we investigated the molecular mechanisms underlying these effects. Western blot analysis demonstrated that RuNC@HS treatment significantly suppressed the protein expression of key components in both the NLRP3/caspase-1 and HMGB1/TLR4/STAT3 pathways in joint tissues (Fig. 7M and N; Fig. S9A–G). These findings suggest that RuNC@HS nanozymes mitigate OA progression by modulating macrophage polarization through inhibition of these pro-inflammatory signaling pathways, thereby disrupting the deleterious crosstalk between BMSCs and macrophages in the osteoarthritic joint. Our integrated analysis reveals that RuNC@HS concurrently modulates the HMGB1/TLR4/STAT3 signaling axis and reshapes PUFA metabolism. While the precise mechanistic crosstalk requires further investigation, we propose a synergistic model wherein these effects are bidirectionally interlinked. On one hand, inhibition of the pro-inflammatory HMGB1/TLR4/STAT3 pathway likely reduces the transcriptional activation of inflammatory enzymes such as COX-2 and 5-LOX, thereby diminishing the consumption of arachidonic acid (AA) for pro-inflammatory eicosanoid production. This suppression at the signaling level contributes to the observed metabolic shift in PUFA composition. On the other hand, the RuNC@HS-driven restoration of FADS2 function and the resultant elevation of ω-3 PUFAs (EPA/DHA) are not merely passive consequences. These ω-3 derivatives serve as precursors for specialized pro-resolving mediators (SPMs), which are known to actively suppress TLR4-mediated signaling and NLRP3 inflammasome activation [75,76]. Thus, the metabolic reprogramming toward a pro-resolving lipid milieu may create a positive feedback loop that further dampens HMGB1/TLR4/STAT3 pathway activity. Collectively, RuNC@HS appears to disrupt a vicious inflammatory cycle by dual mechanisms: directly inhibiting the HMGB1/TLR4/STAT3 signaling cascade while simultaneously promoting a pro-resolving metabolic landscape. The altered PUFA profile thus represents both a consequence of reduced inflammatory drive and an active contributor to the sustained anti-inflammatory phenotype. Future studies employing lipidomic profiling and genetic models will be crucial to fully elucidate the temporal and causal relationships within this intricate network. Translational outlook and challenges of RuNC@HS While the therapeutic efficacy of RuNC@HS is promising, its clinical translation necessitates careful consideration of scalability, regulatory pathways, and positioning within the competitive landscape of nanozyme therapeutics. Firstly, the scalable production and batch consistency of RuNC@HS present a key challenge. Our current synthetic protocol, involving mesoporous silica templating and high-temperature pyrolysis, requires precise control over parameters such as Ru precursor concentration, pyrolysis temperature, and the honeysuckle flavonoid loading ratio to ensure reproducible formation of the active sub-nanometric Ru clusters. Future efforts will need to integrate Process Analytical Technology (PAT) to monitor these critical quality attributes in real-time, alongside establishing rigorous batch-release criteria based on hydrodynamic size, zeta potential, and catalytic activity (SOD/CAT units). Secondly, from a regulatory standpoint, RuNC@HS is best classified as a nanobiologic, given its combination of a nanomaterial core (Ru nanoclusters) with a biologically derived coating (honeysuckle flavonoids). This classification places it within a complex and evolving regulatory framework. Demonstrating safety and efficacy to regulatory bodies will require extensive Good Laboratory Practice (GLP) and Good Manufacturing Practice (GMP) studies, including detailed assessments of long-term biodistribution, potential metal ion leaching, and immunogenicity [77]. Finally, it is instructive to compare RuNC@HS with other nanozyme platforms for OA. Iron-based nanozymes often exhibit peroxidase-like activity under acidic conditions, potentially exacerbating oxidative stress [78], while cerium oxide nanozymes' efficacy can be limited by their band gap energy and environmental-dependent redox switching [79]. Gold nanozymes, though highly biocompatible, typically possess weaker intrinsic catalytic activity [80]. In contrast, RuNC@HS demonstrates a key advantage through its balanced and potent dual SOD/CAT-mimetic activity at physiologically relevant pH, enabling efficient ROS scavenging without generating deleterious intermediates. Furthermore, the honeysuckle decoration provides a synergistic immunomodulatory function not inherent to most metallic nanozymes, positioning RuNC@HS as a uniquely multi-functional therapeutic agent for OA. Conclusion In summary, this study successfully designed and synthesized a ruthenium-based nanozyme, ruthenium nanoclusters (RuNCs), which exhibits dual superoxide dismutase (SOD) and catalase (CAT)-mimicking activities. Further functionalization with honeysuckle flavonoids yielded a nanoimmunomodulator, RuNC@HS. Systematic in vitro and in vivo experiments demonstrated that RuNC@HS effectively scavenges reactive oxygen species (ROS), suppresses NLRP3/caspase-1-mediated pyroptosis in bone marrow mesenchymal stem cells (BMSCs), and reduces HMGB1 release, thereby inhibiting the activation of the TLR4/STAT3 pathway in macrophages. This disruption of the maladaptive inflammatory crosstalk between BMSCs and macrophages ultimately promotes cartilage matrix synthesis, restores subchondral bone architecture, modulates the synovial immune microenvironment, and improves joint function. The study pioneers the integration of traditional herbal active components with precise nanozyme design and elucidates the immunometabolic regulatory mechanisms underlying BMSC-macrophage interactions in osteoarthritis (OA), offering a novel therapeutic strategy for OA management. With excellent biocompatibility, potent ROS-scavenging capacity, and remarkable immunomodulatory properties, RuNC@HS holds significant clinical translation potential, particularly as a candidate for intra-articular injection therapy, potentially providing an effective alternative to current anti-inflammatory treatments for advanced OA patients. However, this study has limitations that highlight directions for future development. The scalable production and batch-to-batch consistency of RuNC@HS require further optimization using advanced manufacturing controls. Comprehensive long-term toxicological profiles and biodistribution studies under GLP standards are essential to de-risk clinical translation. Furthermore, the variability in treatment responses across individuals and OA subtypes warrants future investigation into biomarker-guided patient stratification. Finally, as a novel nanobiologic, the regulatory pathway for RuNC@HS must be clearly defined, necessitating direct engagement with regulatory agencies early in the development process. Future research should focus on optimizing targeted delivery systems, evaluating efficacy across OA subtypes, and advancing imaging- and biomarker-guided personalized therapeutic strategies to facilitate the transition to clinical applications. Funding This research was supported, in whole or in part, by Guangdong Provincial Science and Technology Program - International Science and Technology Innovation Center Construction of the Guangdong-Hong Kong-Macao Greater Bay Area (Guangdong-Hong Kong-Macao Joint Innovation Subprogram) (Bin Yu, 2024A0505040014), Guangdong Basic and Applied Basic Research Foundation – Provincial-Enterprise Joint Funding Program (Key Projects) (Bin Yu, 2021B1515230004), National Key Research and Development Program of China (Bin Yu, 2022YFC2504305), the National Natural Science Foundation of China (Zhiqing Li, Grant Number 82472556) and Guangdong Basic and Applied Basic Research Foundation (Zhiqing Li, Grant Number 2024A1515012825, 2025A1515011535). CRediT authorship contribution statement Yuan Ma: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. Shan He: Conceptualization, Data curation, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. Hanzhong Liu: Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing – review & editing. Jinye Zhang: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Visualization, Writing – review & editing. Wenguo Huang: Formal analysis, Investigation, Methodology, Resources, Software, Validation. Yuan Gu: Data curation, Formal analysis, Methodology, Resources, Software. Naiqian Cui: Investigation, Methodology, Resources, Software. Jikun Qian: Resources, Validation, Writing – review & editing. Mincheng Lu: Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing – review & editing. Daorong Xu: Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – review & editing. Zhiqing Li: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Writing – review & editing. Bin Yu: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – review & editing. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments The characterization results were supported by Beijing Zhongke Baice Technology Service., Ltd. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103176. Contributor Information Mincheng Lu, Email: lumincheng@smu.edu.cn. Daorong Xu, Email: joaquin_xu@126.com. Zhiqing Li, Email: lizq@smu.edu.cn. Bin Yu, Email: yubin@smu.edu.cn. Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request. Supplementary data The following is/are the supplementary data to this article. Multimedia component 1 References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.