Adaptable Covalent Organic Framework‐Hydrogel Microneedles for Glucose‐Responsive Isoliquiritigenin Delivery in Diabetic Wound Therapy Authors: Menghan Zhou, Zongjie Hu, Xiuwen Li, Huaman Geng, Yang Liu, Yajie Wang, Yunshu Yang, Guangdong Zhou, Yujie Hua, Di Wang Published: 2026 ABSTRACT ABSTRACTEffective management of diabetic wound healing remains a significant clinical challenge, primarily due to the persistent inflammatory microenvironment and the lack of effective sustained drug delivery strategies that promote tissue regeneration. This pathological stagnation is increasingly attributed to the overactivation of Z‐DNA binding protein 1 (ZBP1), a pivotal innate immune sensor that triggers pro‐inflammatory cascades under hyperglycemic conditions. In this study, we developed an adaptive covalent organic framework–hydrogel microneedle (ISL@bCOF‐MNs) for glucose‐responsive delivery of isoliquiritigenin (ISL). The system utilizes boron–oxygen bonds to conjugate the small‐molecule drug ISL with a covalent organic framework, which is integrated into microneedles composed of a biomimetic interfacial‐bonding nanocomposite hydrogel, which exhibits high transdermal permeability for deep dermal drug delivery. In vitro experiments show that the ISL@bCOF nanodrug dynamically releases ISL in response to hyperglycemic conditions, thereby suppressing inflammation and scavenging reactive oxygen species via targeting of ZBP1. In vivo studies confirm that the microneedle‐based transdermal delivery system effectively regulates inflammation and promotes full‐thickness wound healing in diabetic mouse models. This work elucidates a potential mechanism by which ISL targets ZBP1 to correct the inflammatory microenvironment and enhance collagen tissue regeneration with advanced biomaterials, offering a promising translational strategy for the clinical management of diabetic wounds. Introduction Diabetic wound healing impairment represents a major clinical challenge in diabetes management, frequently resulting in persistent non‐healing wounds, infection‐related sequelae up to and including amputation [1, 2]. Wound healing is a temporal sequence of hemostasis, inflammation, new tissue formation, and tissue remolding [3, 4, 5]. However, the process of diabetic wound healing is continuously impeded by long‐lasting inflammatory responses during the early period, leading to a dilemma of completing later‐stage skin remodeling under a harsh microenvironment [6]. This pathological stagnation is driven by a recalcitrant hyperglycemic microenvironment characterized by the overactivation of innate immune sensing, excessive production of reactive oxygen species (ROS), and a failure of the macrophage‐mediated transition from a pro‐inflammatory to a pro‐healing phenotype [7, 8, 9]. Consequently, restoring the homeostatic balance of the wound milieu to facilitate tissue regeneration remains a paramount clinical objective. The persistent inflammatory arrest in diabetic wounds is increasingly attributed to the overactivation of Z‐DNA binding protein 1 (ZBP1), a pivotal innate immune sensor [10, 11, 12]. In the hyperglycemic microenvironment, aberrant Z‐RNA accumulation triggers ZBP1‐mediated pro‐inflammatory cascades, effectively blocking the transition to tissue regeneration [13, 14, 15]. Consequently, ZBP1‐targeted immunomodulation represents a transformative strategy to correct this dysfunctional milieu. Traditional Chinese Medicine (TCM) monomers offer a robust library of potential ZBP1 modulators [16, 17]. However, their rational identification remains a challenge. By harnessing artificial intelligence (AI) for large‐scale virtual screening, the traditional process of drug discovery can be revolutionized [18, 19, 20, 21]. This study integrates AI‐driven discovery with biochemical validation to identify high‐affinity TCM monomers, bridging the gap between advanced computational screening and precision immunotherapy for chronic diabetic wound management. For drug encapsulation and delivery, nanocarrier‐based systems offer several advantages over direct delivery, including the encapsulation of hydrophobic drugs, the immobilization of biomolecules, and the protection of enzymes [22, 23]. Among numerous nanomaterials, covalent organic frameworks (COFs) are a novel class of porous nanocarriers with high encapsulation ratios, dynamic response properties, and tunable chemical structures, enabling the design of smart‐responsive nanodrug delivery systems adaptable to pathological wound microenvironments [24, 25, 26]. Hydrogel‐based microneedles easily penetrate the outer skin barrier to transdermally deliver drugs directly to the dermis; they have emerged as a promising strategy for diabetic wound treatment [27, 28, 29]. Although conventional hydrogels exhibit good cytocompatibility because of their high water content, they are often limited by weak mechanical strength in microneedle applications. Recently, a biomimetic interfacial‐bonding nanocomposite (BIN) hydrogel, which mimics the soft‐hard interface of natural tissues to provide robust mechanical strength, combining o‐nitrobenzyl alcohol‐terminated tetra‐armed polyethylene glycol (PEGNB) as a soft phase with high‐density methacrylate‐grafted hyaluronic acid (HAMA) as a hard phase, has demonstrated robust mechanical properties and rapid photo‐fabrication during the preparation of highly permeable microneedles [30]. The elaborate design of COF‐based nanodrugs on microneedles is expected to create a transformative strategy in treating diabetic wounds. In this work, we created an adaptive, glucose‐responsive nanodrug‐in‐microneedle system (ISL@bCOF‐MNs) designed to break the ZBP1‐mediated inflammatory stalemate and accelerate diabetic wound regeneration (Figure 1). In this study, RNA sequencing (RNA‐seq) screening found that ZBP1 was involved in activating early inflammatory responses. Inspired by AI‐driven drug discovery, a small‐molecule TCM monomer (isoliquiritigenin, ISL) was optimized because of its high affinity with ZBP1 and efficient anti‐inflammatory and antioxidant functions [31, 32]. To alter the pathological microenvironment, a novel nanodrug delivery system combining boron‐oxygen (B─O) linked ISL with covalent organic framework (ISL@bCOF) was developed to dynamically respond to a high‐glucose microenvironment, suppress inflammation responses, and scavenge reactive oxygen species (ROS) by targeting ZBP1. Additionally, robust microneedles with high transdermal permeability enabled nanodrug encapsulation and intelligent response delivery. In vivo experiments demonstrated that the microneedle‐based nanodrug transdermal delivery strategy effectively regulated inflammatory responses, promoting full‐thickness wound healing in diabetic mouse models. Overall, this work provides a sophisticated translational strategy, integrating AI‐inspired drug selection with advanced materials science, to address the multifaceted challenges of diabetic wound management. Schematic illustration of the fabrication of ISL@bCOF‐MNs for ZBP1‐targeted immunomodulation and their therapeutic mechanism in diabetic wound healing. Results and Discussion AI‐Driven Drug Scanning and Biological Function Evaluation This study aimed to reverse the inflammatory microenvironment and identify the immunotherapeutic target during diabetic wound healing using a multi‐dimensional drug scanning approach (Figure 2a). An in vitro inflammation model was created using lipopolysaccharide (LPS)‐induced RAW264.7 macrophages, and total gene expression was analyzed using RNA‐seq [33, 34]. Volcano plot and heatmap revealed that a total of 3594 genes were significantly differentially expressed (DEGs) between the control and LPS‐stimulated cohorts (Figure 2b,c). Of these, 1558 were upregulated, while 2036 were downregulated (|log2FC| > 1 and p < 0.05). The predominant biological signatures revealed by GO enrichment analysis centered on immune‐related activities. Specifically, the leading 15 signaling pathways and bioprocesses were concentrated within the domains of “immune system process” and “negative regulation of inflammatory response” (Figure 2d). As demonstrated by the above transcriptome analysis, ZBP1 was found to be significantly differentially expressed (p < 0.01) with a log2FC value of 3.96. ZBP1 exhibited a substantial increase in gene expression in response to LPS stimulation, indicating that it could potentially serve as an important target in regulating the inflammatory microenvironment. Evaluations of AI‐driven ZBP1‐targeted drug screening. (a) Schematic diagram of screening dimensions for ZBP1‐targeted Traditional Chinese Medicine (TCM) monomers. (b–d) Volcano plot (b), heatmap (c), and gene ontology enrichment chart (GO, d) comparing LPS‐induced macrophages with untreated macrophages. (e) Binding patterns and energies of the top ten TCM monomers targeting the ZBP1 protein. (f) CCK‐8 assay was performed on the top ten TCM monomers at 10 µm. (g, h) Representative fluorescence images of RAW264.7 cells stained with M1 and M2 macrophage markers. M1: CD86 (g) or M2: CD206 (h), green; DAPI, blue; Phalloidin, red. (i–k) The semi‐quantitative statistical data of CD86 (i), CD206 (j), and reactive oxygen species (ROS, k) relative fluorescence intensity for each group. (l) Comprehensive evaluation of binding energy, hydrogen bonds, cytocompatibility, anti‐inflammatory, and antioxidant capacity of the ten TCM monomers. Data are presented as mean ± s.d., n = 3 independent replicates. Statistical analysis was performed using one‐way analysis of variance (ANOVA) combined with Tukey's multiple‐comparison test. Differences between groups indicated by different lowercase letters are statistically significant (p < 0.05). The corresponding full names and abbreviations of the ten TCM monomers are listed in Figure S1. To further identify drugs targeting ZBP1, we screened TCM monomers that could modulate inflammatory responses from over a thousand monomers in the Chinese Medicine Monomer Comprehensive Resource Database (http://www.tcmip.cn/ETCM/) using high‐throughput artificial intelligence (AI)‐ driven drug discovery [35]. Specifically, this AI‐driven virtual screening was conducted through a two‐stage process. In the first stage, we searched the database and identified 10 candidate TCM monomers. In the second stage, molecular docking simulations were performed using AutoDock software to evaluate the binding affinities of these 10 TCM monomers with the ZBP1 protein conformation. Autodock molecular docking results illustrated the top ten TCM monomers targeting ZBP1 protein: shikonin, veratric acid, luteolin (LUT), rosmarinic acid (RA), emodin (EM), puerarin (Pue), astragalus polysacharin (APS), mangiferin (Man), isoliquiritigenin (ISL), and quercetin hydrate (QCT) (Figure 2e and Figure S1) [36]. The top five binding energies were APS (−3.31 kcal/mol), EM (−3.24 kcal/mol), Pue (−3.02 kcal/mol), ISL (−2.82 kcal/mol), and LUT (−2.56 kcal/mol), whle the top five hydrogen bonds were Man (seven), ISL (five), LUT (five), RA (four), and QCT (four). To evaluate the cytocompatibility of TCM monomers, a cell counting kit (CCK)‐8 assay was conducted using the RAW264.7 macrophage model with 10 µm for all monomers, showing that SK, LUT, EM, Man, ISL, and QCT were not significantly toxic to macrophages (Figure 2f). To investigate morphological changes, M1 and M2 macrophages were identified by CD86 (Figure 2g) and CD206 labeling (Figure 2h), respectively, and then treated with ten TCM monomers. Fluorescence microscopy imaging showed stronger fluorescence intensity for the M2‐type CD206 marker and weaker intensity for the M1‐type CD86 marker in the LPS + ISL, LPS + QCT, LPS + LUT, and LPS + RA groups compared with the findings in the LPS group (Figure 2i,j). We also evaluated the anti‐oxidant levels of the ten TCM monomers, and intracellular ROS levels were assessed in LPS‐induced macrophages using a DCFH‐DA (1:1000) probe (Figure S2). Following a semi‐quantitative analysis of ROS fluorescence intensity, Pue, APS, QCT, and ISL were identified as potent TCM monomers for effectively scavenging ROS (Figure 2k). Following comprehensive evaluations of binding energy, hydrogen bonds, cytocompatibility, anti‐inflammation, and anti‐oxidant capacity of the ten TCM monomers (Figure 2l), ISL was selected for subsequent experiments because of its overall performance. Synthesis and Characterization of ISL@bCOF Nanodrugs To enhance the drug loading capacity and smart‐response delivery of isoliquiritigenin (ISL), we selected a COF nanocarrier using a solvent‐assisted post‐synthetic modification method [37]. The abundant hydroxyl (─OH) functional groups within the COF structure serve as covalent anchoring sites, enabling the formation of boron‐oxygen (B─O) linkages between ISL and COF architecture (Figure 3a) [38]. The synthetic route of complexation combining boric acid with ISL and its reversible decomposition by glucose is shown in Figure S3. First, scanning electron microscopy (SEM) analysis of ISL@bCOF nanoparticles demonstrated a densely cross‐linked mesh characteristic of a coral‐like structure, while maintaining overall porosity (Figure 3b). Such a conjugation strategy underpins the superior drug‐loading capacity and structural stability validated in our subsequent characterizations. The crystalline integrity of ISL@COF was further substantiated by high‐resolution transmission electron microscopy (HRTEM), which showcased distinct lattice fringes with a d‐spacing of approximately 0.25 nm, consistent with the (100) crystallographic planes (Figure 3c,d). A ring pattern was observed in the selected area electron diffraction. The d‐spacings calculated from this pattern correlated well with those measured from the HRTEM images (Figure 3e). Scanning TEM in dark field with EDS analysis also clarified that C, N, O, and B elements were equably allocated in ISL@bCOF nanoparticles (Figure 3f). Additionally, powder x‐ray diffraction analysis was employed in the low‐angle region. An apparent diffraction peak at 2θ = 3.2° was evident (Figure 3g), confirming the COF crystalline structure. Upon drug loading, the disappearance of the low‐angle diffraction peaks indicated complete pore filling by ISL drugs. Synthesis and characterization of ISL@bCOF nanodrugs. (a) Synthesis route of boron‐oxygen‐linked ISL and covalent organic framework (ISL@bCOF). (b) Representative scanning electron microscopy (SEM) images of ISL@bCOF nanoparticles. (c) Representative high‐resolution transmission electron microscopy (HRTEM) image of ISL@bCOF nanoparticles. (d) The enlarged image of the pink frame in Figure 3c. The insets display the corresponding Fast Fourier Transform (FFT) patterns, while the lattice structure analysis involves a sequential process of FFT, digital filtering, and Inverse FFT (IFFT) reconstruction, supplemented by a quantitative line profile to determine the lattice spacing. (e) The corresponding selected area electron diffraction image was taken from the ISL@bCOF nanoparticles. (f) Elemental mapping of C, N, O, and B by SEM examination. (g) Powder x‐ray diffraction patterns of COF and ISL@bCOF nanoparticles, as well as ISL‐released bCOF nanoparticles. (h, i) Attenuated total reflection Fourier transform infrared (ATR‐FTIR) spectra of synthesizing COF (h) and ISL@bCOF (i) nanoparticles. (j) ATR‐FTIR spectra showed the release of ISL from ISL@bCOF nanoparticles in PBS solution (pH = 5.0). (k) Thermogravimetric analysis (TGA) of COF and ISL@bCOF nanoparticles. (l) The release profile of ISL from ISL@bCOF in glucose solution and PBS. Data are presented as mean ± s.d., n = 3 independent replicates. (m) Nitrogen adsorption isotherm and desorption isotherm for COF (open symbols) and ISL@bCOF (filled symbols) nanoparticles. (n) The dV/dD pore volume of COF and ISL@bCOF nanoparticles. To substantiate the effective encapsulation of ISL, ATR‐FTIR spectroscopy was employed to elucidate the chemical integrity of the system. The complete disappearance of characteristic peaks at 3345 and 3280 cm− 1 (N─H stretching vibrations of amine groups) and at 1643 cm− 1 (C═O stretching vibration of aldehyde monomers) served as a confirmation of quantitative conversion during the polycondensation reaction (Figure 3h) [39, 40]. The emerging vibrational bands at 1594 (C═C stretching) and 1271 cm− 1 (C─N stretching) provided definitive evidence for the formation of β‐ketoenamine linkages in the COF structure [41]. ATR‐FTIR spectroscopy confirmed the structural integrity of COF after ISL loading, with the C═C (1594 cm− 1) and C─N (1271 cm− 1) stretching vibrations remaining stable (Figure 3i). The C─H stretching vibrations of trimethyl borate at 2956 and 2872 cm− 1 significantly diminished, while a new characteristic peak emerged at ∼1050 cm− 1, corresponding to the B─O─C bond formation through transesterification [42]. Additionally, the appearance of distinctive ISL vibrational bands at 1146 and 1167 cm− 1 further corroborated successful drug loading. To evaluate the drug release behavior of ISL@bCOF, the composite was dispersed in phosphate‐buffered saline (PBS) buffer (measured pH = 5.0) under continuous agitation for 72 h. The disappearance of both the B─O─C bond vibration (∼1050 cm− 1) and characteristic ISL peaks (1146 and 1167 cm− 1) confirmed the complete release of ISL from ISL@bCOF (Figure 3j). The fundamental mechanism driving this glucose‐responsive delivery relies on the dynamic covalent nature of the B─O─C bond. Glucose molecules contain multiple cis‐diols, which have a much higher binding affinity for boronic acid than the hydroxyl groups of ISL [43]. Under hyperglycemic conditions typical of diabetes, diffusing glucose molecules act as competitive triggers. By competitively binding to the boronic acid sites on the COF framework, they initiate an ester exchange reaction [44, 45]. This exchange rapidly cleaves the original B─O─C bond anchoring the ISL, leading to the structural dissociation of the drug complex and, consequently, the rapid release of the ISL. Thermal decomposition profiles acquired in a nitrogen environment revealed that the initial weight reduction for both pristine COF and ISL@bCOF was negligible (< 5%) prior to reaching 150°C, indicating superior thermal integrity (Figure 3k). ISL@bCOF exhibited enhanced thermal stability in comparison with the unloaded COF (350°C), a phenomenon attributable to the integration of robust B─O covalent linkages. Following the application of a temperature of 400°C, a consistent mass difference of approximately 30% was observed between the two materials, corresponding to the ISL loading capacity. This result aligns well with the theoretical drug‐feeding ratio, indicating efficient covalent conjugation rather than physical adsorption. Full‐spectrum ISL showed a maximum absorption wavelength of approximately 370 nm (Figure S4). Compared to PBS, the in vitro release capacity of ISL loaded on ISL@bCOF was significantly enhanced in 50 µm glucose solution (Figure 3l). These findings not only verify that the ISL@bCOF nanocarrier system enables a sustained and controlled release profile of the encapsulated ISL but also demonstrate that the presence of glucose can effectively facilitate the drug release process of this system. To investigate the surface morphology and internal porosity, nitrogen sorption experiments were performed at liquid nitrogen temperature. The pristine COF exhibited a specific surface area of 131.6 m2 g− 1, with pore size analysis revealing a primary distribution around 2.1 nm. The specific surface area of ISL@bCOF was 12.5 m2 g−1, which was smaller than that of the original COF (Figure 3m). Figure 3n depicted that the dV/dD pore volume of ISL@bCOF was 0.0015 cm3 g−1, which was smaller than that of the original COF material (0.025 cm3 g−1). Taken together, these results demonstrated that ISL@bCOF nanodrugs were successfully prepared, showing a typical crystalline architecture, enhanced drug loading efficiency, and smart‐response delivery characteristics for diabetic wound healing. Fabrication and Characterization of ISL@bCOF‐MNs Patch To engineer a robust microneedle (rMN) patch for ISL@bCOF nanodrug encapsulation and transdermal delivery, a PEGNB/HAMA hydrogel system was selected [30]. In this system, a novel phototriggered transient‐radical and persistent‐radical coupling reaction was developed to provide robust mechanical properties and rapid photo‐fabrication, facilitating ISL@bCOF nanodrug incorporation into hydrogel needles to prepare the ISL@bCOF‐MNs patch. For puncturing performance in transdermal delivery applications, a high‐density cross‐linked HAMA (molecular weight 44 kDa, grafting degree of ∼95%) was chosen with a PEGNB (40 kDa, labeling rate approximately 3.8–4.0) component to enhance microneedle rigidity. As shown in Figure 4a, the gel precursor of PEGNB/HAMA (7% w/v) mixed with ISL@bCOF nanodrugs was poured into microneedle molds and quickly yielded mechanically robust hydrogel arrays upon light irradiation (365‐nm LED, 20 mW/cm2). Figure 4b depicted the general morphology of a square microneedle with a side length of 1 × 1 cm, which displayed 12 × 12 arrays in a rectangular pyramid with dimensions of 400 × 400 × 800 µm (W × L × H). To thoroughly characterize microneedle morphology, cryogenic scanning electron microscopy (cryo‐SEM) was employed to capture both top‐view and 45° tilted‐view images of ISL@bCOF‐MNs arrays. High‐magnification photography of an individual needle tip showed a well‐aligned microstructure with intact and sharp geometry (Figure 4c). Complementary characterization using confocal laser scanning microscopy (CLSM) with rhodamine B‐labeled microneedle tips further confirmed the cryo‐SEM structural observations (Figure 4d). The fabrication and characterization of the ISL@bCOF‐MNs patch. (a) Schematic of ISL@bCOF‐MNs patch fabrication with sequential mold casting and rapid photo‐fabrication. (b) Representative photographs of the ISL@bCOF‐MNs patch. (c) Representative cryogenic scanning electron microscopy (cryo‐SEM) images of the bottom (i) and side (ii, iii) views of the ISL@bCOF‐MNs patch. (d) Representative confocal laser scanning microscopy (CLSM) of 2D (i) and 3D (ii) images showing the rhodamine B‐labeled ISL@bCOF‐MNs patch. (e) Rheological analysis of PEGNB/HAMA and HAMA hydrogels (G’: storage modulus; G”: loss modulus). (f–h) Representative tensile (f) and compressive (g) stress–strain curves, as well as Young's modulus (h) of PEGNB/HAMA and HAMA hydrogels. (i) Release profiles of ISL drugs from ISL@bCOF‐MNs patch in PBS with different glucose concentrations (0, 5, 10, and 15 mm). A two‐way analysis of variance (ANOVA) combined with Tukey's multiple‐comparison test was performed; significant differences compared to the 0 mm control group were noted only on day 16. (j) Representative photographs of MNs loaded with indocyanine green (ICG) and fluorescence imaging of ICG‐loaded MNs after transdermal administration in porcine skin at days 1, 3, and 5. (k) Relative fluorescence mean of transdermal ICG delivery on porcine skin at days 1, 3, and 5. (l) Live/dead cell staining of fibroblasts after co‐culturing with different MN extracts for 48 h. (m) Hemolysis ratios of different MN extracts (PC: ddH2O; NC: PBS) (n = 6). Data are presented as mean ± s.d., n = 3 independent replicates. * p < 0.05, *** p < 0.001, and **** p < 0.0001. Hydrogel microneedles require robust mechanical strength to achieve deep transdermal penetration. To evaluate the intrinsic macroscopic mechanical properties of the hydrogel matrices, tensile and compressive tests were first performed on bulk dumbbell‐shaped and cylindrical samples, respectively. The gelation time of PEGNB/HAMA hydrogels was determined to be approximately 8 s when exposed to 365‐nm LED light irradiation (Figure 4e). In comparison with HAMA hydrogels alone, PEGNB/HAMA hydrogels exhibited higher tensile strength, ductility, and toughness (Figure 4f–h). The tensile curve of PEGNB/HAMA hydrogels showed nearly 1.5‐fold tensile strain reaching approximately 4.2 MPa tensile strength because of their mechanical enhancement of biomimetic soft‐hard material interface microstructures, whereas HAMA hydrogels displayed weak tensile characteristics. Similarly, PEGNB/HAMA hydrogels showed nearly 5 MPa compressive strength, 5‐fold higher than that of HAMA hydrogels. In addition, PEGNB/HAMA hydrogels showed a nearly 6‐fold increase in Young's modulus compared with HAMA hydrogels. Furthermore, to evaluate the mechanical structural integrity under practical application scenarios, compression tests were conducted simultaneously on fully formed microneedle patches containing multiple tips. The resulting force‐displacement curve (Figure S5) demonstrated that the microneedle array possesses sufficient mechanical robustness to withstand skin penetration without buckling, highlighting its suitability for transdermal drug delivery. The release kinetics of bioactive substances from hydrogel microneedles play a crucial role in diabetic wound healing. The boron‐oxygen bonds formed in the ISL@bCOF structure are liable to undergo responsive rupture under specific high‐glucose conditions, leading to a controlled release of their cargo. The release characteristics of ISL from ISL@bCOF‐MNs were assessed at varying glucose concentrations (0, 5, 10, and 15 mm). There was a quicker release of ISL at higher glucose concentrations (10 and 15 mm) compared with that at 0 mm glucose (Figure 4i). The results demonstrate that ISL@bCOF‐MNs possess a glucose‐responsive release of ISL, demonstrating an intelligent response to the diabetic microenvironment. The transdermal delivery ability of MNs was assessed by observing the fluorescence intensity of indocyanine green (ICG) with an in vivo imaging system. MNs were loaded with ICG and placed into pig skin, with fluorescence intensity detected after 1, 3, and 5 days. Progressive fluorescence was noted in the skin over 5 days, indicating successful transdermal permeation and sustained retention of ICG beyond the stratum corneum (Figure 4j,k). This indicated that the MNs have sufficient mechanical strength and puncture performance for transdermal delivery applications. Immunoregulatory Mechanism of ISL@bCOF Nanodrugs Involves Targeting ZBP1 An in vitro inflammation model was established using lipopolysaccharide (LPS) induced macrophages, and differentially expressed genes (DEGs) were analyzed by RNA‐seq, revealing that the immunomodulatory mechanism of ISL@bCOF depends on ZBP1. As demonstrated by the above transcriptome analysis, ZBP1 was found to be significantly differentially expressed (p < 0.01) with a log2FC value of 3.96. ZBP1 exhibited a substantial increase in gene expression in response to LPS stimulation, indicating that it could potentially serve as an important target in regulating the inflammatory microenvironment. Transcriptomic analysis identified 52 DEGs (|log2FC| > 1, p < 0.05) between the LPS and LPS+ISL@bCOF groups, comprising 15 upregulated and 37 downregulated genes (Figure 5a). Gene ontology enrichment analysis was performed to elucidate the functional landscape of the DEGs, revealing a predominant distribution in critical biological processes such as cellular, metabolic processes, and immune system processes (Figure S6). Further pathway analysis using the KEGG and GO enrichment analysis of the top 15 biological processes revealed significant modulation of processes including “inflammatory bowel disease”, “positive regulation of inflammatory response”, etc (Figure 5b,c). Notably, gene set enrichment analysis (GSEA) of these immune‐related pathways, coupled with the corresponding expression heatmap, pinpointed ZBP1 as a central node. ISL@bCOF treatment potently suppressed LPS‐induced upregulation of ZBP1 (Figure 5d–f). Collectively, these results implicate ZBP1 as a pivotal target in modulating the inflammatory microenvironment, suggesting its critical role in the mechanism through which ISL@bCOF promotes diabetic wound healing. Underlying immunoregulatory mechanism of ISL@bCOF nanodrugs. (a) Volcano plot of ISL@bCOF nanodrug treatment in LPS‐induced macrophages. (b, c) KEGG enrichment (b) and GO enrichment (c) of ISL@bCOF nanodrug treatment in LPS‐induced macrophages. (d, e) Gene set enrichment analysis (GSEA) for the positive regulation of inflammatory response (d) and inflammatory response (e). (f) Heatmap for the corresponding pathway in Figure 5e. (g) Protein‐protein interaction (PPI) network constructed via STRING and visualized in Cytoscape. (h) Down‐expressed genes in the KEGG enrichment analysis of ISL@bCOF nanodrug treatment in LPS‐induced macrophages. (i) Molecular docking of ISL with the ZBP1 protein. (j) Binding affinity and kinetics evaluation between ISL and ZBP1 protein. The solid black lines represent the 1:1 binding fit, while the colored lines represent the experimental data. Protein‐protein interaction (PPI) network analysis identified a highly interconnected inflammatory hub, where IL‐1β and IL‐1α serve as central nodes with robust functional ties to the innate immune sensor ZBP1, suggesting a coordinated regulatory axis in the inflammatory response (Figure 5g). Furthermore, Figure 5h indicates that down‐expressed genes in the KEGG enrichment analysis of ISL@bCOF nanodrug treatment in LPS‐induced macrophages. In addition, we tested ISL and ZBP1 molecular docking interactions using AutoDock, showing that ISL could bind to histidine 155 and glutamate 122 in ZBP1 with relatively high affinity (−2.82 kcal/mol) (Figure 5i). To validate the direct interaction between ISL and ZBP1, we performed surface plasmon resonance (SPR)‐based kinetic analysis [46]. As shown in Figure 5j, ISL exhibited a clear concentration‐dependent binding response to the immobilized ZBP1 protein. The binding profiles were characterized by rapid association and relatively fast dissociation phases, typical of small‐molecule interactions. The experimental sensorgrams were well‐described by the 1:1 Langmuir binding model, as evidenced by the high quality of the fit with a low value of 0.112. Kinetic analysis revealed an association rate constant (ka ) of 3.43 × 103 M− 1s− 1 and a dissociation rate constant (kd ) of 8.24 × 10− 2 s− 1. The calculated equilibrium dissociation constant (KD ) was 24.1 µm, indicating a moderate and specific binding affinity between ISL and ZBP1. These results demonstrate that ISL is a direct binder of the ZBP1 protein. The expression distribution of key inflammatory mediators (IL‐1β, IL‐6, IL‐10) and ZBP1 among the DEGs was also quantified (Figure S7). In vitro Validation of ZBP1 Regulation in the Inflammatory Immune Microenvironment To systematically evaluate the biocompatibility of ISL and ISL@COF, a CCK‐8 assay analysis was conducted utilising the RAW264.7 macrophage model. As demonstrated in Figure 6a, when ISL concentration was maintained at or below 10 µm, the cell viability remained above 95% (p > 0.05 vs. Control), thereby substantiating this concentration as the maximum non‐toxic threshold. Figure 6b demonstrated that ISL@COF exhibited superior biocompatibility compared to ISL, with an increased safety concentration up to 40 µm. Calculated at a drug loading rate of 30%, the effective concentration of ISL is 12 µm. This is attributed to the sustained‐release properties of the COF, which reduced the burst toxicity of the drug. Molecular validation of ZBP1 regulation of inflammatory responses in vitro experiments. (a, b) CCK‐8 assay was performed with different concentrations of (a) ISL (10 µm) and (b) ISL@COF (40 µm) (n = 6). (c) Flow cytometry analysis of CD206 and CD86 expression for the control, LPS, LPS + ISL@bCOF, LPS + ISL@bCOF + ZBP1, and LPS + ISL@bCOF + CsA groups. CsA: ZBP1 protein inhibitor. (d, e) Statistical histograms of M1 (d) and M2 (e) macrophages in each group. (f) Reactive oxygen species (ROS, green) fluorescence images with different treatments. (g) Relative mRNA expression levels of IL‐1β, IL‐6, IL‐10, and ZBP1 in each group. (h, i) Western blot analyses of protein expression levels of IL‐1β, IL‐6, IL‐10, and ZBP1 in each group. (j) Schematic of the immunoregulatory mechanism of ISL@bCOF nanodrugs targeting ZBP1 protein. Data are presented as mean ± s.d., n = 3 independent replicates. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001. On this basis, we hypothesized that the ISL@bCOF nanodrug exerted anti‐inflammation and antioxidation effects by mediating macrophage phenotypic polarization via targeting ZBP1, while concomitantly scavenging intracellular ROS. To verify this hypothesis, further molecular biology assessments were performed by adding an activator (ZBP1) or an inhibitor (cyclosporin A [CsA]) with ISL@bCOF nanodrugs [47]. An LPS‐induced macrophage model was utilized to investigate ISL@bCOF functions, and the resulting polarization efficacy was analyzed with flow cytometry. The experiments were divided into five groups, including the control, LPS, LPS + ISL@bCOF, LPS + ISL@bCOF + ZBP1, and LPS + ISL@bCOF + CsA. The polarization effects of COF and ISL were analyzed separately (Figure S8). Relative to the control, LPS stimulation significantly increased the proportion of M1 (CD86+) macrophages. Conversely, ISL@bCOF treatment not only reversed this effect by reducing M1 polarization but also markedly promoted the polarization toward the M2 (CD206+) phenotype (Figure 6c). ZBP1 addition inhibited the anti‐inflammatory effect of ISL@bCOF nanodrugs, while the addition of CsA synergistically enhanced their M2‐type macrophage polarization efficacy. Quantitative data further confirmed that more M1‐type macrophages were converted to the M2 anti‐inflammatory phenotype because of the effective inhibition of ZBP1 gene expression, implying the suppression of ZBP1‐dependent inflammation responses (Figure 6d,e). The dysfunctional healing microenvironment in diabetic wounds is characterized by a persistent inflammatory state, compounded by markedly elevated oxidative stress [48]. To further evaluate oxidative stress, intracellular ROS levels were assessed using the DCFH‐DA (1:1000) probe in LPS‐induced macrophages treated with ISL@bCOF, ISL@bCOF + ZBP1, and ISL@bCOF + CsA. ISL@bCOF nanodrugs exhibited ROS scavenging capacity, which was attributed to the boron‐oxygen bond between ISL and COF nanoparticles (Figure 6f and Figure S9). The mRNA expression levels of M1 macrophage‐associated inflammatory factors, M2‐associated anti‐inflammatory factors, and ZBP1 were quantified using quantitative real‐time polymerase chain reaction (qRT‐PCR) (Figure 6g) [49, 50]. The effects of COF and ISL on the mRNA expression levels of M1 macrophage‐associated inflammatory factors (IL‐1β, TGF‐β1, and iNOS) and M2‐associated anti‐inflammatory factor (IL‐10) are shown in Figure S10. Western blot analysis of protein expression yielded consistent results with qRT‐PCR (Figure S11). Compared with ISL@bCOF treatment alone, ISL@bCOF treatment with the addition of ZBP1 upregulated ZBP1 gene expression while downregulating the anti‐inflammatory cytokine IL‐10 and upregulating pro‐inflammatory factors IL‐1β and IL‐6; in contrast, ISL@bCOF treatment with the addition of CsA showed opposite regulatory effects. Western blot analysis of ZBP1 protein expression yielded consistent results with qRT‐PCR (Figure 6h,i). The LPS + ISL@bCOF + ZBP1 group showed significantly increased protein expression of IL‐1β, IL‐6, and ZBP1 compared with the ISL@bCOF group, along with decreased IL‐10 protein expression. Conversely, LPS + ISL@bCOF + CsA further reduced the protein expression levels of IL‐1β, IL‐6, and ZBP1 while increasing IL‐10 expression. The relative fluorescence intensity was significantly higher in the LPS + ISL@bCOF + ZBP1 group compared with the intensities in the LPS + ISL@bCOF and LPS + ISL@bCOF + CsA groups. This was likely because of the activation of ZBP1, weakening the ROS scavenging ability of ISL@bCOF nanodrugs, while its inhibition led to enhanced ROS scavenging capability. Taken together, ISL@bCOF nanodrugs showed an immunoregulatory role with ZBP1‐dependent anti‐inflammation and antioxidation effects (Figure 6j and Figure S12), modulating the pathological microenvironment in diabetic wound healing. In vivo Therapeutic Performance of Microneedles on Diabetic Diabetic Wounds To assess the therapeutic potential of the ISL@bCOF‐MNs patch on diabetic wound healing, full‐thickness cutaneous wounds (6 mm in diameter) were created on db/db mice (C57BL/6J background) to establish a diabetic mouse model with impaired wound healing. To evaluate the therapeutic efficacy of the ISL@bCOF‐MNs patch as a wound dressing, we employed this animal model. Mice were randomly allocated into four experimental groups: control, ISL@bCOF, MNs, and ISL@bCOF‐MNs. The ISL@bCOF‐MNs patch significantly accelerated wound closure after 14 days of treatment, with complete scab detachment observed, whereas scabs remained visible in other groups (Figure 7a,b). By day 16, the wounds of both the control and MNs groups were still unhealed; however, complete wound healing had occurred in the ISL@bCOF treated group, and the wounds in the ISL@bCOF‐MNs group had significantly healed. For clearer visualization of wound healing rates across four groups, the healing area trajectories at different time points were analyzed by ImageJ (Figure 7c,d). These results demonstrated that the ISL@bCOF nanodrug‐treated group exhibited increased diabetic wound healing through the immunoregulatory effect of ISL release, while the ISL@bCOF‐MNs patch group demonstrated enhanced efficacy in accelerating wound healing because of the wet healing environment of the rMN patch dressing. The blood glucose of mice in the four groups showed a high concentration over 16 days (Figure 7e), indicating the long‐lasting inflammatory environment of diabetic wounds and continuous glucose‐responsive ISL cargo release. Therapeutic effects of microneedles on diabetic wounds in diabetic mouse models. (a) Schematic of ISL@bCOF‐MNs patch treatment on wounds in diabetic mouse models. (b) Representative photographs of the diabetic wounds for different groups (ISL@bCOF, MNs, ISL@bCOF‐MNs, and control) on days 0, 3, 5, 7, 10, 14, and 16. (c, d) Area diagrams (c) and the corresponding statistical analysis (d) of diabetic wound healing over time. (e) Histological examinations of H&E and Masson's trichrome staining of the diabetic wounds in different groups on day 16. Blue arrows represent the edges of wounds. Red boxes represent the interfacial regions; green boxes represent the repaired regions. Data are presented as mean ± s.d., n = 3 independent replicates. *** p < 0.001, and **** p < 0.0001. The evaluation of the therapeutic effect of ISL@bCOF‐MNs on diabetic wound healing was performed with hematoxylin‐eosin (H&E) staining of skin tissues on day 16. Panoramic views of different skin tissues were taken along with magnified images of two distinct regions (red and green boxes), with blue arrows representing wound boundaries (Figure 7f). Both ISL@bCOF and ISL@bCOF‐MNs groups exhibited more intricate, re‐epithelialized structures, with epithelialization being more pronounced in the ISL@bCOF‐MNs group. However, these structures were absent in the control and MNs groups, and epidermal healing was incomplete in the control group. By day 16, Masson's trichrome staining revealed substantial extracellular matrix (ECM) deposition in both ISL@bCOF and ISL@bCOF‐MNs groups, suggesting that collagen in the newly formed tissues was comparable to that of normal tissues, with ECM deposition being more pronounced in the ISL@bCOF‐MNs group. Conversely, the control and MNs groups exhibited minimal ECM deposition, with the ECM distribution being sparse and irregular at day 16. Additionally, the global histological landscape provided by Sirius Red staining underscored the enhanced structural integrity of the nascent dermis in ISL@bCOF‐MNs treated wounds, characterized by dense, well‐aligned collagen fibers that effectively bridged the lateral wound margins (Figure 8a). Sirius Red staining showed that the ISL@bCOF‐MNs group presented strong red and green polarized light signals with significantly higher brightness than the other three groups, suggesting that type I and type III collagen fibers were abundant and in a coordinated ratio. H&E, Masson's trichrome, and Sirius Red staining revealed that the ISL@bCOF‐MNs patch significantly accelerated the regenerative repair and remodeling of diabetic wounds. Vascularization, defined as the growth of new blood vessels and cell proliferation, is a significant indicator of skin regeneration in the context of diabetic wound healing. The histogram of collagen angle distribution derived from Sirius Red‐stained polarized light results reveals that the control group, ISL@bCOF group, and MNs group exhibit dominant orientation near 0° (frequency >20%), while the ISL@bCOF‐MNs group shows a significant increase in orientation frequency at −50° (exceeding 30%) and a reduced proportion at 0° orientation (Figure 8b). The corresponding polar plot further validated this trend: compared to the other groups concentrated in the central region, the ISL@bCOF‐MNs group exhibited a significant shift toward the −60° direction (Figure 8c). To rigorously decouple the therapeutic contributions of the nanocarrier and the microneedle platform, an additional control group consisting of free ISL loaded directly into microneedles without the bCOF carrier (ISL@MNs) was evaluated (Figures S13). Histological analysis at Day 16 revealed that the absence of the glucose‐responsive bCOF carrier compromised long‐term tissue remodeling. Unlike the highly aligned collagen network promoted by ISL@bCOF‐MNs, the Sirius Red polarized light analysis of the ISL@MNs group demonstrated a random and disorganized distribution of collagen fibers. These decoupled findings firmly validate that the integration of the bCOF nanocarrier is essential for providing the sustained, microenvironment‐responsive drug release necessary for high‐quality chronic wound regeneration. Orientation and distribution of collagen in diabetic wound healing. (a) Sirius red staining of the diabetic wounds in each group on day 16. (b, c) Histogram analysis (b) and polar coordinate plot analysis (c) of collagen fiber angle distribution in dorsal skin tissue from mice in each group. Inflammation and Tissue Remodeling These observations indicate that ISL@bCOF‐MNs induced specific reorientation of the surface structure, potentially providing a structural basis for regulating macrophage polarization and ROS clearance. Angiogenesis and cell proliferation were assessed on day 16 via immunofluorescence staining for CD31 and Ki67, respectively (Figure 9a). Representative images and subsequent quantification demonstrated that, notably, both ISL@bCOF and ISL@bCOF‐MNs treatments markedly promoted capillary regeneration and increased the proliferation rate relative to control conditions (Figure 9d,e). The ISL@bCOF‐MNs group exhibited the best neovascular density and cell proliferation, implying favorable wound healing induced by the microneedle‐based nanodrug transdermal delivery strategy. Immunofluorescence evaluation of diabetic wound healing and immunoregulatory effects. (a) Immunofluorescence staining of Ki67 (red) and CD31 (green) in different macrophages from the diabetic mouse model treatment groups (ISL@bCOF, MNs, ISL@bCOF‐MNs, and control) on day 16 after treatment. Nuclei were stained with DAPI (blue). (b, c) Immunofluorescence staining of CD86 (red, b) and CD206 (red, c) markers in different groups after 3 and 7 days of treatment. Nuclei are stained with DAPI (blue). (d) Quantification of cell proliferation in each group using Ki67 staining. (e) Quantification of capillary density in each group using CD31 staining. (f, g) Quantification of the percentage of CD86 (f) and CD206 (g) immune cells in each group after 3 and 7 days of treatment. White arrows represent blood vessels. Red boxes represent the interfacial regions; green boxes represent the repaired regions. Data are presented as mean ± s.d., n = 3 independent replicates. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001. The immunofluorescence staining results of M1‐type macrophages (CD86+ CD68+) and M2‐type macrophages (CD206+ CD68+) on days 3 and 7 show that ISL@bCOF and ISL@bCOF‐MNs groups had lower proportions of M1 macrophages at day 3 (Figure 9b,c). By day 7, the ISL@bCOF group still contained a small number of M1 macrophages in the dermis, while the ISL@bCOF‐MNs group demonstrated comprehensive anti‐inflammatory effects in both epidermal and dermal layers. In contrast, the control and MNs groups maintained significant inflammatory signals (p < 0.01). The quantitative analysis of mean values associated with CD86+ and CD206+ expression is shown in Figure 9f,g, respectively. While transdermal delivery of free ISL provided a transient anti‐inflammatory response during the early healing phase (Day 3), it failed to sustain M2 macrophage polarization through Day 7 due to rapid drug depletion (Figures S14) [51]. The immunoregulatory performance of the ISL@bCOF‐MNs patch may be attributed not only to continuous drug release from ISL@bCOF prodrugs but also to the microneedle‐based transdermal penetration, enabling deeper and more precise drug delivery. Conclusions This study presents a novel ZBP1‐targeted immunotherapeutic strategy for chronic wound healing, addressing key challenges through the following innovative components. By establishing an AI‐driven TCM monomer screening method, ISL was identified as an effective ZBP1‐targeted agent, accelerating precise drug discovery. The development of COF‐based nanodrugs (ISL@bCOF) with smart, glucose‐responsive delivery ensures targeted anti‐inflammatory and anti‐oxidant effects while reducing biosafety risks. Additionally, BIN hydrogel MNs provide efficient transdermal penetration and a physical barrier, enhancing drug delivery in harsh wound microenvironments. Verified in diabetic mouse models, this approach highlights the critical role of ZBP1 in early immunoregulation for subsequent tissue remodeling. Though challenges such as large animal trials and standardized production remain, these results advance AI‐assisted nanomaterial‐based immunotherapy, offering a versatile platform for inflammation‐related regenerative repair and paving the way for clinical translation in chronic wound management. Experimental Section Materials Shikonin, veratric acid, LUT, RA, EM, Pue, APS, Man, ISL, and QCT were purchased from Shanghai Yuanye Bio‐Technology Co (Shanghai, China). Ltd. 3,3'‐Dihydroxybenzidine, 2,4,6‐triformylphloroglucinol, 1,2,4,5‐benzenetetramine tetrahydrochloride, polyphosphoric acid, 1,4‐dioxane, glacial acetic acid, methyl alcohol, trimethyl borate, and anhydrous acetonitrile were purchased from Shanghai Titan Technology Co (Shanghai, China). LPS, Triton X‐100, DAPI, and BSA were purchased from Beijing Solarbio Technology Co., Ltd (Beijing, China). The ROS probe (DCFH‐DA) was purchased from Shanghai Biyuntian Biotechnology Co., Ltd. The abbreviations involved in this study are summarized in Table S1 for clarity. Cell Culture Fibroblasts and Raw264.7 cells were cultured in high‐glucose Dulbecco's modified Eagle's medium (DMEM, Gibco, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, USA) and 1% penicillin‐streptomycin. All cells were maintained at 37°C in a humidified incubator with 5% CO2. Eukaryotic Sibling Transcriptome Sequencing The RNA libraries were sequenced by OE Biotech, Inc. (Shanghai, China). Bioinformatic analysis was performed using OECloud tools at https://cloud.oebiotech.com/task/. The volcano plots and other graphics were drawn using R (https://www.r‐project.org/) on the OECloud platform (https://cloud.oebiotech.com/task/). Synthesis of COF The target COF was synthesized via a solvothermal method. An amino monomer (50 mg, 231 µmol) and 1,3,5‐triformylphloroglucinol (32.4 mg, 154 µmol) were charged into a Pyrex tube, followed by the addition of trimethylbenzene (1.5 mL), dioxane (1.5 mL), and 6 m acetic acid (0.5 mL). The resulting mixture was sonicated for 10 min to ensure thorough homogenization. The Pyrex tube was then rapidly frozen in liquid nitrogen, evacuated to remove dissolved gases from the solution, and thawed to room temperature. This freeze–pump–thaw cycle was repeated three times to ensure complete removal of oxygen. Subsequently, the tube was sealed and heated in an oven to 120°C at a rate of 5°C min− 1 and maintained at this temperature for 72 h. After the reaction was complete, the tube was allowed to cool naturally to room temperature. The resulting red‐orange precipitate was collected by filtration using a 0.22 µm organic membrane. The obtained solid was further purified by Soxhlet extraction with methanol for 24 h. Finally, the product was dried under vacuum at 60°C for 12 h to afford the final COF product. Synthesis of ISL@bCOF In a glove box, ISL (30 mg), COF (60 mg), and trimethyl borate (10.4 mg, 100 µm) were added to a 10 mL tube with 3.5 mL anhydrous acetonitrile. The mixture was stirred at 60°C for 4 h under argon, then heated to 70°C, nitrogen‐purged to remove by‐product methanol, and reacted for 1 more day (with methanol removed regularly). After cooling, the product was transferred to a flask, solvent‐evaporated, vacuum‐dried at 60°C for 12 h, and stored in a glove box. The coupling between COF and ISL was achieved via a transesterification reaction. Specifically, trimethyl borate was introduced as a linker, which has been reported to undergo transesterification with phenolic hydroxyl groups [52, 53]. Trimethyl borate provides multiple reactive sites that can interact with the hydroxyl groups of the COF building units as well as the multiple hydroxyl sites of ISL, thereby enabling effective grafting. Analytical Techniques Microstructural morphology of samples was observed by employing the ZEISS Sigma 300 scanning electron microscopy (SEM), and Cryogenic electron microscopy images were investigated by Talos F200X G2 (HRTEM). Powder x‐ray diffraction patterns were recorded using a Bruker D8 Advance diffractometer. All samples were recorded in the 2–40° (2θ) range at 0.02° increments. Attenuated total reflectance Fourier‐transform infrared spectroscopy (ATR‐FTIR) spectra were taken on a Nicolet 6700. Thermogravimetric analysis (TGA) curves were recorded on a PerkinElmer TGA8000. Nitrogen sorption isotherms were measured with a Micromeritics ASAP 2020 PLUS HD88 Microporous physical adsorption apparatus. The Brunauer–Emmett–Teller (BET) method was utilized to calculate the specific surface areas. Synthesis of ISL@bCOF‐MNs PEGNB/HAMA hydrogel (G001) was purchased from Shanghai Tiling Medical Technology Co. The MN models were filled with 120 µL of ISL@bCOF‐PEGNB/HAMA hydrogel and then centrifuged at 2500 xg for 5 min to ensure that the solution filled the top of the model without air bubbles. The MN patch was irradiated with UV light for 2 min to cure, and this was removed to obtain ISL@bCOF‐MNs. The same procedure was followed for control MNs preparation. Surface Plasmon Resonance Analysis The binding kinetics and affinity between isoliquiritigenin (ISL) and ZBP1 protein were characterized using a Biacore 8K molecular interaction system (Cytiva). The ZBP1 protein, diluted to 25 µg/mL in 10 mm sodium acetate buffer (pH 4.0), was covalently immobilized onto a CM5 sensor chip via standard amine coupling chemistry. The immobilization procedure included surface activation for 7–10 min, ligand coupling for 1200 s, and final quenching with 1 m ethanolamine for 7 min. For kinetic measurements, ISL was prepared in a series of concentrations (100, 50, 25, 12.5, and 6.25 µm) using PBST buffer supplemented with 5% DMSO. A multi‐cycle kinetics method was employed with a contact time of 120 s and a dissociation time of 120 s at a constant flow rate of 30 µL/min. Solvent correction was performed to account for the refractive index changes caused by DMSO. All sensorgrams were processed using Biacore Insight Evaluation Software (version 6.0.7). The experimental data were fitted to a global 1:1 binding model to determine the association rate constant (ka ), dissociation rate constant (kd ), and equilibrium dissociation constant (KD ) [46]. Cytocompatibility Cells were inoculated in 96‐well plates at a density of 5000 cells per well and treated with drugs at specific ratios. An aliquot of 10 µL of CCK‐8 reagent (CK04, Dojindo, Shanghai, China) was added to each well and incubated for 2 h away from light before absorbance was measured at 450 nm. Other CCK‐8 assay steps were as described above. Fibroblasts were inoculated into ISL@bCOF hydrogel‐impregnated high‐sugar medium for 3 days and then incubated in Petri dishes for 2 days. A Calcein‐AM/PI (live/death) staining kit (C542, Dojindo, Shanghai, China) was used to detect cell activity according to the kit instructions. Haemolysis Assay Collect fresh rabbit blood into EDTA anticoagulant tubes; add PBS, centrifuge at 1500 rpm for 10 min, and discard the supernatant. Repeat the above steps until the PBS supernatant becomes colorless, yielding a thoroughly washed erythrocyte pellet. Dilute with PBS to obtain a 2% red blood cell suspension. Experimental groups included positive control, negative control, MNs, and ISL@bCOF‐MNs groups. Slowly aspirate 5 µL of the 2% red blood cell suspension from the bottom of the anticoagulant tube into a 1.5 mL centrifuge tube. Add 1 mL of the corresponding microneedle's three‐day extraction solution to each experimental group. Add 1 mL of PBS to the negative control group and 1 mL of ddH2O to the positive control group. After thorough mixing of the solutions with the red blood cell pellet, incubate all groups at 37°C in a constant‐temperature incubator for 2 h. After incubation, centrifuge at 1500 rpm for 3 min. Observe color differences between the supernatant of each group and the positive/negative controls. Transfer the supernatant from each group to a 96‐well plate. Calculate the hemolysis rate by measuring the absorbance at 540 nm. Sustained‐Release Experiment The prepared ISL@bCOF‐MNs were suspended in 5 mL of PBS and divided into five groups with different glucose concentrations (0, 5, 10, and 15 mm). All samples were incubated in a 37°C constant‐temperature shaking incubator. On days 1, 3, 5, 7, 10, 14, and 16, 3 mL of supernatant was collected from each group. Absorbance values (OD) were measured at the peak wavelength of 370 nm. After each sampling, 3 mL of PBS corresponding to the respective glucose concentration was replenished into each group. Samples were then returned to the 37°C shaking incubator for continued agitation until the next sampling and analysis. Anti‐Inflammatory Assay RAW264.7 cells (density: 5 × 106 cells) were inoculated into six‐well plates. After cell attachment, the cells were then co‐treated with 100 ng/ml LPS and ISL@bCOF for 24 h. ZBP1 (1:1000, HY‐P71436, MCE, Shanghai, China) and cyclosporin A (1:2000, HY‐B0579, MCE, Shanghai, China) were added to the mixed solution proportionally in separate experiments. After cells were collected, they were incubated with antibodies CD86 and CD206, respectively, and analyzed with flow cytometry. Antioxidant Assay RAW264.7 cells were inoculated in 6‐well plates (1 × 106 cells/well). The cells were stimulated with 100 ng/ml LPS to induce oxidative stress and then treated with ISL@bCOF, ISL@bCOF + ZBP1, or ISL@bCOF + CsA for 24 h. DCFH‐DA (1:1000) (S0033S, Biyuntian, Shanghai, China) was added and allowed to react at 37°C for 0.5 h. The fluorescence intensity was observed under a Leica (DMI8) microscope and quantified using ImageJ. Animal Experiment A total of 36 male db/db mice (C57BL/6J background, 6–8 weeks) were used. Mice were depilated on the back with an electric hair clipper and depilatory cream. Full skin trauma (6 × 6 mm) was caused by excising skin from the back. The mice were randomly allocated to four groups and treated for 16 days with ISL@bCOF, MNs, or ISL@bCOF‐MNs. The control received no treatment. Samples were taken on days 3, 7, and 16 for subsequent experiments. (1) woundhealingrate%=1−A0At×100% where A0 and At represent the wound area at the initial and predetermined time points, respectively. qRT‐PCR and Western Blotting Analysis Total RNA was extracted using chloroform according to previously described methods and reverse‐transcribed using a Takara kit [54]. Gene quantification was performed using SYBR Green. Quantitative analysis of mRNA expression for inflammation‐related genes (IL‐1β, IL‐6, IL‐10, and ZBP1) was performed by qRT‐PCR using the 2−ΔΔCT method, with β‐actin serving as the endogenous control. Primer sequences are detailed in Table S2. Protein analysis was conducted via Western blotting. Total protein was extracted using RIPA lysis buffer (P0013B, BiyunTian, China), separated by SDS‐PAGE, transferred to a membrane, and probed with specific antibodies for detection. Histological and Immunohistochemical Analyses Specimens were first fixed in 4% paraformaldehyde for 1 day before paraffin‐embedding and sectioning into 4 µm thick sections. Sections were further stained with H&E, Masson's trichrome, and Sirius red to assess the regenerated tissue structure according to the manufacturer's protocol. Tissue sections were stained by immunofluorescence with anti‐CD31 (1:500) and anti‐Ki67 (1:100) antibodies to detect traumatic angiogenesis and cell proliferation, respectively. Macrophage polarization was detected by applying anti‐CD68 (1:50), anti‐CD86 (1:200), and anti‐CD206 (1:200) antibodies. Information on antibodies used for immunofluorescence staining is listed in Table S3. Statistical Analysis All quantitative data were expressed as mean ± s.d., n = 3 independent replicates. Statistical analysis and graphing were performed using GraphPad Prism 9.5.0 and Origin 2024. For statistical comparisons between two independent groups, an unpaired two‐tailed Student's t‐test was used. For datasets involving two independent variables, two‐way analysis of variance (ANOVA) was performed, followed by Tukey's post hoc test. In specific multi‐group evaluations, statistical significance may also be indicated by a letter‐coding system, where groups not sharing a common letter are significantly different from one another. A p‐value < 0.05 was considered significant for all statistical tests. Ethics Statement All protocols for experiments involving animals were approved by the Animal Care and Experimental Committee of Shandong Second Medical University (approval number: 2024SDL270). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supporting File: adhm71343‐sup‐0001‐SuppMat.docx.