Bio-inspired supramolecular hydrogel inhibits ferroptosis to accelerate diabetic wound healing

Graphical abstract

Highlights

• •High-glucose microenvironment induces endothelial cell ferroptosis.
• •Baicalein suppresses endothelial cell ferroptosis through modulation of the Keap1/NRF2/HIF-1α signaling pathway.
• •Bai@GA/Gel/Zn²⁺ supramolecular nanofibrous hydrogel enables sustained baicalein release and demonstrates multifunctional capacities, including promoting cellular proliferation, exerting antibacterial effects, and inhibiting ferroptosis.
• •Bai@GA/Gel/Zn²⁺ exhibits remarkable therapeutic efficacy in diabetic chronic wound treatment.

Abstract

Introduction

Ferroptosis significantly compromises diabetic wound healing and represents a promising therapeutic target for clinical intervention. Baicalein (Bai), a natural flavonoid, has been considered to own ferroptosis-regulating properties, while the underlying mechanisms remain understood.

Objectives

The study aims to investigate potential mechanisms by which Bai inhibits ferroptosis in endothelial cells and validate the therapeutic efficacy of Bai-loaded hydrogel on diabetic wounds.

Methods

High-glucose-cultured human umbilical vein endothelial cells (HUVECs) were chosen to study Bai’s molecular mechanisms inhibiting ferroptosis. Subsequently, a Bai@GA/Gel/Zn²⁺ supramolecular nanofibrous hydrogel was developed for active Bai delivery. Its wound-healing efficacy and underlying mechanisms were further evaluated in a diabetic rat model.

Results

In vitro experiments, HUVECs were observed to exhibit characteristic ferroptosis, which were induced by high-glucose (HG) microenvironment, leading to compromised cellular functions. Through the inhibition of ferroptosis in HUVECs, Bai effectively promoted cell growth, migration, and angiogenesis. Herein, we further reveal that Bai inhibits HG-induced ferroptosis by modulating the Keap1/NRF2/HIF-1α signaling axis. To optimize therapeutic delivery, Bai was encapsulated into supramolecular nanofibers self-assembled from glycyrrhizic acid (GA), followed by cross-linking with gelatin (Gel) and zinc ions (Zn²⁺) to form a hydrogel (Bai@GA/Gel/Zn²⁺), thereby enabling sustained release of Bai at the wound-healing interface. Bai@GA/Gel/Zn²⁺ nanofiber-incorporated hydrogel demonstrated favorable mechanical and biological properties, markedly reducing ferroptosis levels and accelerating diabetic wound healing.

Conclusion

In summary, our findings reveal the ferroptosis-inhibiting mechanism of Bai and develop a Bai@GA/Gel/Zn²⁺ supramolecular nanofibrous hydrogel that synergistically enhances diabetic wound healing with proven therapeutic efficacy.

Introduction

Diabetic chronic wounds, driven by metabolic dysregulation and pathological mechanisms such as persistent inflammation and vascular dysfunction, exhibit severely impaired angiogenesis due to microcirculatory failure [1], [2], [3], [4], [5]. Ferroptosis, a regulated process of cell death marked by reactive oxygen species (ROS) buildup, lipid peroxidation, and disrupted iron metabolism, is a critical contributor to impaired diabetic wound healing [6], [7]. Therapeutic strategies targeting ferroptosis offer promising potential for restoring endothelial function and promoting vascular regeneration [8], [9], [10].

Plant defense mechanisms inspire novel diabetic wound strategies. Baicalein (Bai), a Scutellaria baicalensis flavonoid, exhibits potent anti-inflammatory and antibacterial properties. Its unique 5,6,7-trihydroxyflavone structure provides exceptional antioxidant capacity to alleviate oxidative stress and crucially suppress ferroptosis [11], [12], a pivotal process in impaired diabetic healing. To effectively deliver Bai and create a synergistic therapeutic environment, we designed a supramolecular hydrogel using glycyrrhizic acid (GA) as a multifunctional scaffold. This natural triterpenoid self-assembles into nanofibers that significantly enhance the solubility of the poorly water-soluble Bai, effectively encapsulating it for sustained release [13], [14] and mimic the extracellular matrix to support cellular adhesion, proliferation, and enhanced drug-cell interactions [14], [15].

Herein, we engineered a supramolecular nanofibrous hydrogel by integrating Bai into self-assembled GA nanofibers and subsequently complexing with gelatin (Gel) and zinc ions (Zn²⁺), the latter being strategically selected not only as a crosslinker that forms metal-coordination bonds with GA to enhance mechanical strength but also as an intrinsic antibacterial agent. The resulting structure is stabilized through this dual physical crosslinking, which involves metal-coordination bonds between Zn²⁺ and GA as well as hydrogen bonds between Gel and GA nanofibers. This rationally designed network significantly imparts greater mechanical strength and stability to the hydrogel while enabling sustained drug release (Fig. 1). Within this system, Bai reduces oxidative stress and regulates iron metabolism to protect endothelial cells from ferroptosis. The antibacterial action of Bai and GA is further potentiated by the incorporated Zn²⁺, creating a synergistic effect that disrupts bacterial cell walls and inhibits metabolic enzymes. In addition, self-assembled GA nanofibers mimic collagen morphology to modulate cell migration, promote vascularization, and suppress inflammatory responses. Subsequently, in vivo experiments validated the capacity of the Bai@GA/Gel/Zn²⁺ hydrogel to accelerate wound healing through enhanced collagen deposition, revascularization, and ferroptosis suppression, demonstrating considerable promise for clinical applicability.

Fig. 1.

Schematic illustrations of the fabrication process and therapeutic mechanism for accelerating diabetic wound healing of Bai@GA/Gel/Zn²⁺ nanofiber hydrogel.

Materials and methods

Materials and instruments

Drugs including Bai, RSL-3, Ferrostatin-1 (Fer-1), Gel, and GA were sourced from Aladdin in Shanghai, China; and Cell Counting Kit-8 (CCK-8) from Beyotime (Shanghai, China), as well as Malondialdehyde Assay (MDA), ROS Assay and Live/Dead Cell Staining Kit. The Ferrous Iron (Fe²⁺) Detection Kit was secured from Elabscience (Wuhan, China). Primary antibodies against Nuclear factor erythroid 2 related factor 2 (NRF2), Beta-Actin (β-actin), Tumor Necrosis Factor α (TNF-α), Nuclear Receptor Coactivator 4 (NCOA4), and Mouse IgG were supplied by Proteintech (Wuhan, China). Antibodies targeting Hypoxia-inducible factor 1 alpha (HIF-1α), Glutathione Peroxidase 4 (GPX4), Transferrin receptor protein 1 (TFR1), and Kelch-like ECH-associated protein 1 (Keap1) were provided by Abmart (Shanghai, China). The experiments were performed using instruments from the following suppliers: chemiluminescence imaging system (Tanon, China); electrophoresis apparatus (Bio-Rad, USA); CO₂ incubator and biological safety cabinet (Thermo Fisher, USA); inverted biological microscope (Leica, Germany); multimode microplate reader (Tecan, Switzerland); and flow cytometer (BD Biosciences, USA).

Fabrication and characterization of Bai@GA nanofibers and Bai@GA/Gel/Zn²⁺ nanofibrous hydrogels

Bai@GA nanofibers were fabricated via the self-assembly of GA in aqueous solution. Briefly, GA (80 mg) and Bai (8 mg) were dissolved in 3 mL of anhydrous ethanol under low-power ultrasonication, thus adding deionized water (10 mL). Subsequently, dialysis of the mixture was performed in deionized water with a 3.5 kDa MWCO under light protection for 24 h to remove free drugs and organic solvents. The clear nanofiber supernatant was then collected by centrifugation (4000 rpm), and the nanofibers’microstructural geometry was analyzed utilizing transmission electron microscopy (TEM).

To prepare Bai@GA/Gel/Zn²⁺ hydrogels, Gel and Zn²⁺ were separately dissolved in deionized water for 30% (w/v) Gel and 1% (w/v) Zn²⁺ stock solutions, which were mixed with the Bai@GA nanofiber suspension at 40°C in specific ratios to reach end concentrations at 0.2%, 0.4%, or 0.6% (w/v) for Bai@GA, 10% (w/v) for Gel, and 0.1% (w/v) for Zn²⁺. The homogeneous mixture was transferred into molds and gelled at 4°C overnight. Lyophilized hydrogels were examined for internal microstructure via scanning electron microscopy (SEM). Dynamic rheology was measured using an MCR301 rheometer.

Adhesion and Elasticity of Bai@GA/Gel/Zn²⁺

The adhesive properties of the hydrogel patch were evaluated by attaching it to a finger joint and observing its stability during joint flexion. Elasticity was tested by compressing the hydrogel and monitoring its ability to regain its original shape. To evaluate the viscoelasticity of the hydrogel, uniaxial compression and lap shear tests were performed on a universal testing machine. The load and displacement were continuously monitored during testing, from which the elastic and viscous contributions were quantitatively determined.

Swelling, degradation and drug release profiles of Bai@GA/Gel/Zn²⁺

The swelling, degradation, and drug release behaviors of the Bai@GA/Gel/Zn²⁺ hydrogel were systematically investigated under two PH conditions (7.4 and 5.5) to mimic the environments of normal and diabetic wounds, respectively.

For swelling tests, hydrogels were placed in PBS at 37℃ and weighed at predetermined time points. The swelling ratio (S) was calculated as S (%) = ((W_t − W₀)/W₀) × 100%, where W₀ and W_t are the initial and time-dependent wet weights, respectively.

Degradation was monitored by measuring the mass loss of lyophilized hydrogels over one week. The degradation rate (D) was determined as D (%) = [(W₀ − W_d)/W₀] × 100%, with W₀ and W_d denoting the initial dry weight and the dry weight on day d, respectively.

Drug release profiles were established by immersing hydrogels in PBS under continuous shaking (100 rpm) at 37°C. At designated intervals, 1 mL of release medium was collected and replaced with an equal volume of fresh PBS. The concentration of released Bai was quantified using UV–Vis spectrophotometry, and the cumulative release percentage was calculated based on the Lambert-Beer law.

Cells culture

Primary HUVECs were routinely propagated in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS) with 5% CO₂ (37°C temperature). Actively proliferating cells at exponential growth phase were enzymatically detached for downstream experimental processing.

Lentiviral Transduction

HUVECs was transduced with NRF2-targeting or control shRNA lentivirus (GenePharma, China) for 24 h, followed by replacement with fresh medium for another day. Selection was then performed using 1 μg/mL puromycin for 48 h to establish stable knockdown pools. For long-term maintenance, Cell culture was conducted under puromycin selection (0.5 μg/mL).

Biocompatibility and biosafety assessment of Bai@GA/Gel/Zn²⁺

The biosafety of Bai@GA and Bai@GA/Gel/Zn²⁺ was systematically evaluated through a battery of in vitro and in vivo evaluations. In vitro cytotoxicity was assessed using CCK-8 and Live/Dead staining on HUVECs exposed to material extracts. Hemocompatibility was verified by a standard hemolysis assay. Furthermore, to evaluate systemic biosafety, major organs (heart, liver, spleen, lungs, and kidneys) harvested from treated rats were processed for H&E staining.

Cell migration assays

Scratch assay: HUVECs (1 × 10⁶ cells/well) were grown to confluence in 6-well plates. Uniform scratches were made using a sterile 200 μL pipette tip perpendicular to pre-marked grid lines (spaced 0.5–1 cm apart). After PBS washes, the cells were exposed to a serum-free medium, either containing the drug or not. Scratch closure was observed and imaged at both 0 and 24 h, and the mean area of the scratch was quantified via Image J software.

Transwell assay: The upper compartments of 8 μm Transwell inserts were occupied by HUVECs in a serum-free medium at 2 × 10⁴ cells per well, and the lower compartment was filled with compete medium. Following a 24-hour incubation at 37℃, cells remaining in the upper section were carefully wiped away using cotton buds. Cells positioned on the lower membrane, after migration, were stabilized with 4% paraformaldehyde (PFA) solution (15 min) and cellular architectures were visualized through immersion in 1% crystal violet working solution (30 min), and quantified under an optical microscope.

Angiogenesis assay

A total of 50 μL Matrigel (BD Biosciences) was evenly coated onto the bottom of 24-well plates and solidified at 37℃ for 40 min. HUVECs (2 × 10⁴ cells/well) were seeded onto the Matrigel-coated plates and exposed to experimental treatments. Cells underwent staining with Calcein-AM following a 5-hour incubation at 37℃ and imaged using an inverted fluorescence microscope.

Western blotting

HUVEC-derived proteins processed in diverse conditions utilizing RIPA lysis buffer constituted the experimental material in this study, which were then denatured in loading buffer (100℃, 5 min), and quantified via bicinchoninic acid (BCA) assay. After being resolved by SDS-PAGE and electrophoretic transfer onto PVDF membranes, the immobilized proteins were subjected to a blocking procedure using 5% (w/v) non-fat milk solution for 120 min (RT). The membrane specimens were thus immersed in chilled primary antibody diluent (GPX4, NCOA4, TFR1, NRF2, Keap1, β-actin), rinsed in TBST, and exposed to a goat anti-rabbit secondary antibody (1 h, RT). After additional washes with TBST, protein bands were revealed through enhanced chemiluminescence and analyzed densitometrically with ImageJ software.

Quantitative Real-Time PCR (qPCR)

After total RNA extraction and cDNA synthesis, qPCR was performed using SYBR Green Master Mix. Gene expression was quantified via the 2^–ΔΔCt method with β-actin normalization. The primer sequences were:

Primer		Sequences(5'-3')
HIF-1α (human)	Forward	5′-ATCCATGTGACCATGAGGAAATG-3′
HIF-1α(human)	Reverse	5′-TCGGCTAGTTAGGGTACACTTC-3′
NRF2(human)	Forward	5′-AACCAGTGGATCTGCCAACT-3′
NRF2(human)	Reverse	5′-AACGTAGCCGAAGAAACCTCAT-3′
β-actin(human)	Forward	5′-TCACCATGGATGATGATATCGC-3′
β-actin(human)	Reverse	5′-GGAATCCTTCTGACCCATGC-3′

Co-immunoprecipitation (Co-IP)

Cells were lysed and centrifuged to collect supernatant. After quantifying protein concentration by BCA assay, an input sample was reserved. The remaining lysate was pre-cleared with Protein A/G beads (MCE, America), incubated with anti-Keap1 or control IgG antibody for 2 h at 4°C, and then bound to fresh beads overnight at 4°C. The beads were rinsed five times, and the bound proteins were released by boiling in loading buffer (5 min) for subsequent WB analysis.

Intracellular ROS evaluation

To evaluate ROS production, a fluorescence-based detection protocol was implemented. Cultured HUVECs in 12-well plates were processed as designated and subjected to 10 μm DCFH-DA (1:1000) under light protection at 37°C for 30 min. Following enzymatic detachment, cell suspensions were centrifuged, rinsed with PBS buffer and reconstituted in 500 μL of PBS to facilitate subsequent assessment. Fluorescence intensity was quantified using flow cytometry (FITC channel) with FlowJo 10.8 software.

Quantification of MDA and Fe²⁺ levels

HUVECs cultured in 6-well plates were exposed to experimental groups. Intracellular MDA (lipid peroxidation marker) and Fe²⁺ concentration quantification was performed with standardized commercial kits following manufacturer-specified procedures.

In vitro antibacterial assay

The antibacterial efficacy was evaluated against Staphylococcus aureus (S.aureus) and Escherichia coli (E.coli). Bacterial suspensions were co-cultured with test extracts for 12 h at 37°C. Bacterial growth was quantified by measuring OD600 using a microplate reader. Serial dilutions of bacterial cultures were plated on agar, incubated overnight, and colony-forming units (CFUs) were counted and compared across groups.

Diabetic rat wound healing

Male Sprague-Dawley (SD) rats received 65 mg/kg streptozotocin (STZ) through intraperitoneal delivery to induce diabetes. Circular skin wounds with a full thickness and a diameter of 1 cm were surgically removed from the back. Wounds were topically treated with PBS, Bai, Bai@GA, GA/Gel/Zn²⁺ or Bai@GA/Gel/Zn²⁺. Wound healing progression was serially assessed by photographic documentation on days 0, 3, 5, 7, and 9 after the wound occurred. Peri-wound specimens were collected, chemically stabilized using 4% PFA (24 h, RT), and processed for histopathological evaluation, employing dual routine stains (HE and Masson's trichrome) complemented by immunohistochemical (IHC) marker detection (TNF-α, HIF-1α, GPX4, NCOA4 and TFR1), as well as immunofluorescence (IF) staining (CD31 and α-SMA).

Ethic Approval and Consent to Participate

The Institutional Animal Use and Care Committee of the Nanjing Durm Tower Hospital consented to all experiments on animals (Ethics Approval Number: 2025AE02005).

Statistical analysis

Each of the experiments was conducted independently no fewer than three times. Data was expressed as mean ± standard deviation. Data were analyzed by Student’s t-test, one-way ANOVA, or two-way ANOVA in GraphPad Prism 10.0, and differences were regarded as statistically significant at P < 0.05. Graphical representations were generated by Origin 10.1.0.

Results and discussion

High-Glucose Microenvironment Induces Ferroptosis and Impairs Cellular Functions in HUVECs

Endothelial dysfunction is a critical factor contributing to impaired revascularization in diabetic wounds. To investigate whether HG induces ferroptosis in endothelial cells, HUVECs were exposed to different glucose concentrations (5 mM, NG and 33 mM, HG) for 48 h. Functional assays revealed distinct impairments under HG conditions compared to NG controls. CCK-8 analysis demonstrated significantly reduced cell viability in the HG group (Fig. 2a), while scratch closure and Transwell migration tests showed impaired wound closure and fewer migrated cells, respectively, indicating compromised migratory capacity (Fig. 2b-d, S1). Furthermore, tube formation assays highlighted diminished angiogenic potential in HG conditions, characterized by simplified vascular networks with shortened branch lengths (Fig. S2c-d). As depicted in Fig. 2e, the mechanism underlying HG-induced ferroptosis is delineated. Specifically, HG exposure downregulated expression of GPX4, a key ferroptosis suppressor responsible for mitigating oxidative stress (Fig. 2f-g), and intracellular ROS levels were markedly elevated in the HG group (Fig. 2h-i). Concurrently, upregulation of iron metabolism-related genes, including NCOA4 and TFR1, was observed alongside intracellular Fe²⁺ accumulation (Fig. 2k). The HG group also displayed significantly increased MDA levels (Fig. 2j), a hallmark of ferroptosis, indicating enhanced lipid peroxidation. Collectively, these results indicate that HG predispose HUVECs to ferroptosis through amplified oxidative stress and iron dysregulation, disrupting endothelial function and revascularization.

Fig. 2.

High-glucose microenvironment induces ferroptosis and impairs cellular functions in HUVECs. (a) Cell viability of HUVECs. n = 6. (b,d) Images and quantitation of scratch migration in HUVECs. Scale bar is 100  μm. n = 3. (c) Quantitation of Transwell assays in HUVECs. n = 3. (e) Mechanism schematic of high glucose-induced ferroptosis. (f-g) Western blot images and relative quantification of GPX4, NCOA4, and TFR1. n = 3. (h-i) Intracellular levels of ROS in HUVECs. n = 3. (j) Levels of MDA in HUVECs. n = 3. (k) Intracellular Fe²⁺ content in HUVECs. n = 3. Data was analysed using Student’s t-test. *p < 0.05, **p < 0.01, ***p < 0.001.

Bai Attenuates HG-Induced ferroptosis and restores HUVECs functions

Bai, a natural compound with antioxidative and iron-chelating properties, inhibits ferroptosis by neutralizing ROS via its hydroxyl groups and forming stable Fe²⁺/Fe³⁺ complexes to block Fenton reactions [16]. To evaluate the safety profile of Bai for therapeutic applications, HUVECs were exposed to escalating concentrations of Bai (0–100 μM) for 48 h. Cytotoxicity assays confirmed no adverse effects at concentrations ≤ 100 μM (Fig. S2a), prompting the selection of 20 μM Bai for subsequent experiments according to previous research. RSL-3, a canonical GPX4 inhibitor that covalently inactivates GPX4 to induce ferroptosis via peroxidation accumulation, was applied at its IC50 (8 μM) to model ferroptosis (Fig. S2b) [17]. We then investigated whether Bai could counteract ferroptosis-driven endothelial dysfunction in HG or RSL-3 groups. The group design logically delineates the pathogenesis and drug mechanism: NG and HG groups established the HG model; HG + Bai demonstrated Bai's efficacy; while the RSL-3 and RSL-3 + Bai groups crucially confirmed that the cytoprotective effect of Bai is specifically mediated through antagonizing ferroptosis. Functional assays revealed that both HG and RSL-3 significantly impaired cellular functions compared to NG controls, whereas Bai treatment largely restored viability, migration, and angiogenic capacity to near-normal levels (Fig. 3a-e, S2c-d), indicating its protective role against ferroptosis-associated damage. At the molecular level, Bai reversed HG- and RSL-3-induced ferroptosis by dual mechanisms. As a pivotal regulator of ferroptosis, GPX4 maintains membrane lipid bilayer integrity by catalyzing glutathione-dependent reduction of lipid peroxides to lipid alcohols [18]. Western blotting revealed that Bai upregulated GPX4 expression, thereby enhancing cellular defenses against lipid peroxidation (Fig. 3f-g). Meanwhile, Bai restored iron homeostasis by suppressing the overexpression of NCOA4 (a key mediator of ferritinophagy) and TFR1 (a regulator of iron uptake), thereby reducing intracellular Fe²⁺ accumulation (Fig. 3f-g, j). Consistent with these findings, Bai attenuated oxidative stress markers, including ROS and MDA in both HG and RSL-3 groups (Fig. 3h-i, k). Totally, these data confirm that Bai preserves critical endothelial functions by targeting ferroptosis induced by HG or RSL-3.

Fig. 3.

Bai Attenuates HG-Induced Ferroptosis and Restores HUVECs Functions. (a) Cell viability of HUVECs. n = 6. (b-e) Characteristic images and statistical evaluation of scratch migration and Transwell assays in HUVECs. Scale bars: 100 μm (d) and 50 μm(e). n = 3. (f-g) WB images and relative quantitative analysis of GPX4, NCOA4, and TFR1. n = 3. (h) Intracellular levels of ROS in HUVECs. n = 3. (i) Levels of MDA in HUVECs. n = 3. (j) Intracellular Fe²⁺ value in HUVECs. n = 3. (k) Quantitative analysis of ROS levels in HUVECs. n = 3. Groups: NG: Normal glucose; HG: High glucose; HG + Bai: HG with Baicalein; RSL-3; RSL-3 + Bai: RSL-3 with Baicalein. Data was analysed using one way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

Bai inhibits endothelial cell ferroptosis by regulating the Keap1/NRF2/HIF-1α signaling pathway

Although we have demonstrated that Bai restores cellular function by suppressing ferroptosis, its precise mechanism remains unclear. NRF2 is a master transcriptional regulator of lipid peroxidation and ferroptosis, negatively modulated by Keap1 [19], [20]. HIF-1α, a key oxygen-sensing transcriptional activator, promotes angiogenesis by stimulating the expression of vascular endothelial growth factor (VEGF) [21], [22]. Emerging evidence indicates NRF2 can not only enhances HIF-1α protein and mRNA expression, but also stabilizes HIF-1α protein under hypoxic conditions, suggesting an undeniable functional crosstalk between these two critical stress-response pathways [23], [24]. We systematically designed NG, HG, HG + Bai, and HG + Bai + ML385 groups, with the pivotal inclusion of the NRF2 inhibitor ML385 to confirm pathway necessity. To investigate this, we assessed Keap1, NRF2, and HIF-1α protein expression via WB. As shown in Fig 4a-b, HG or RSL-3 stimulation significantly elevated Keap1 and HIF-1α levels while reducing NRF2 expression. Fortunately, Bai reversed these effects, downregulating Keap1/HIF-1α and restoring NRF2. To confirm pathway dependency, NRF2 was pharmacologically inhibited by ML385. WB analysis confirmed that NRF2 suppression concurrently decreased HIF-1α expression (Fig 4c-d). Consequently, Bai's anti-ferroptotic efficacy was markedly attenuated, as evidenced by decreased GPX4 levels, upregulated NCOA4 and TFR1 expression, and elevated ROS, MDA, and Fe²⁺ concentrations (Fig 4c-h). This study reveals that HG-induced ferroptosis in HUVECs is linked to the Keap1/NRF2/HIF-1α pathway, wherein NRF2 inhibition and HIF-1α downregulation jointly contribute to endothelial dysfunction and impaired angiogenesis, leading to chronic wounds. Bai restores redox homeostasis and suppressing ferroptosis by activating the Keap1/NRF2/HIF-1α axis.

Fig. 4.

Bai Inhibits Endothelial Cell Ferroptosis by Regulating the Keap1/NRF2/HIF-1α Signaling Pathway. (a-d) WB images and quantitative analysis of NRF2, Keap1, HIF-1α, GPX4, NCOA4, and TFR1. n = 3. (e-f) Flow cytometry image and quantitation of ROS levels in HUVECs. n = 3. (g) Intracellular MDA levels in HUVECs. n = 3. (h) Intracellular Fe²⁺ value in HUVECs. n = 3. Groups: NG: Normal glucose; HG: High glucose; HG + Bai: HG with Baicalein; HG + Bai + ML385: HG with Baicalein and ML385. Data was analysed using one way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

Bai promotes Keap1-NRF2 dissociation and Upregulates HIF-1α expression

Under basal conditions, Keap1 mediates NRF2 ubiquitination and degradation, maintaining low basal NRF2 activity [25]. Oxidative stress triggers NRF2-Keap1 dissociation, promoting nuclear translocation and anti-oxidative response element (ARE) binding in downstream targets genes such as GPX4 to reestablish redox homeostasis [26]. Notably, a functional ARE has been identified and validated in an upstream enhancer region of the HIF1α gene locus, establishing HIF1α as a direct transcriptional target of NRF2 [27]. To investigate whether Bai directly targets Keap1, structure-based molecular docking was performed. The results indicated that Bai binds tightly to the Keap1 Kelch domain with a favorable binding energy of –8.73 kcal/mol, forming three to four hydrogen bonds with key residues, including VAL-420, VAL-512, and VAL-467 (Fig. 5a). Co-IP assays further verified that Bai promotes the dissociation of the Keap1–NRF2 complex (Fig. 5b).

Fig. 5.

Bai Promotes Keap1-NRF2 Dissociation and Upregulates HIF-1α Expression. (a) Molecular docking between Bai and Keap1. (b) Co-IP assay of Keap1 and NRF2 in HUVECs. n = 3. (c-e) Validation of NRF2 knockdown efficiency in HUVECs by WB and qPCR analysis. n = 3. (f) Quantification of HIF-1α mRNA levels. n = 3. (g-h) WB images and quantitative analysis of HIF-1α, NRF2, Keap1, TFR1, NCOA4, and GPX4. n = 3. Data was analysed using one way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

To functionally validate the role of NRF2 in this regulatory axis, we established NRF2-knockdown HUVECs using lentiviral shRNA. Successful knockdown was confirmed by WB and qPCR analyses (Fig. 5c-e). The protective effect of Bai against ferroptosis was markedly attenuated in NRF2-knockdown cells (Fig. 5g-h, S3a-b), suggesting that NRF2 is essential for Bai's protection against ferroptosis. Moreover, NRF2 knockdown led to marked downregulation of HIF-1α expression (Fig. 5f-h), indicating that NRF2 acts as a positive regulator of HIF-1α. This observation is consistent with the reported NRF2-ARE mechanism governing HIF1α transcription, and confirms that Bai enhances HIF-1α expression primarily through NRF2 activation.

Characterization of Bai@GA/Gel/Zn²⁺ supramolecular nanofiber hydrogel

Compared to standard clinical dressings that primarily provide passive coverage, hydrogels offer a dynamic therapeutic platform capable of actively modulating the wound microenvironment. This functionality is rooted in their three-dimensional polymeric network structure, which can be engineered to sustain the release of drugs or cells and directly intervene in regenerative processes [28], [29], [30]. Recent progress in supramolecular self-assembly now enables the dynamic construction of such bioactive ordered structures through non-covalent interactions, opening novel avenues for the next generation of intelligent hydrogel dressings in regenerative medicine [31]. Herein, we successfully developed the nanofibrous hydrogel through the integration of Bai into self-assembled GA nanofibers, followed by complexation with Gel and Zn²⁺. This composite structure, stabilized by dual physical crosslinking via Zn²⁺-GA coordination and Gel-GA hydrogen bonding, exhibited significantly enhanced mechanical strength, stability, and sustained drug release profile. The hydrogel exhibited an orange hue that intensified with increasing Bai@GA concentration (Fig. 6a-b). TEM confirmed the self-assembled nanofibrous structure of Bai@GA (Fig. 6d), while SEM revealed a porous scaffold in Bai@GA/Gel/Zn²⁺ with embedded nanofibers (Fig. 6c).

Fig. 6.

Characterization of Bai@GA/Gel/Zn²⁺ Supramolecular Nanofiber Hydrogel. (a-b) The Schematic diagram and image of Bai@GA/Gel/Zn²⁺. (c) SEM images of Bai@GA/Gel/Zn²⁺. Scale bars: 100  μm and 50  μm. (d) TEM images of Bai@GA nanofibers. Scale bars: 200  μm. (e-f) FTIR spectra of GA, Bai, Bai@GA, Gel and Bai@GA/Gel/Zn²⁺. (g-h) Rheological frequency and strain sweep of Bai@GA/Gel/Zn²⁺. (i) Live and dead staining of HUVECs. Scale bar: 100 μm. n = 3. (j) Representative images of S. aureus and E. coli after incubation. n = 3. (k) Hemolysis image and rate of Bai@GA/Gel/Zn²⁺. (l) Survival rate of S. aureus and E. coli. n = 3. (m-n) ROS levels and quantitative analysis in HUVECs. n = 3. (o) MDA levels in HUVECs. n = 3. (p) Fe²⁺ value in HUVECs. n = 3. Data was analysed using one way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

FTIR analysis verified the synthesis of Bai@GA and Bai@GA/Gel/Zn²⁺ by identifying characteristic functional group vibrations (Fig. 6e-f). GA displayed characteristic peaks at 3393.23 cm⁻¹ (COOH), 2948.65 cm⁻¹ (C–H), 1716.99 cm⁻¹ (C=O), 1653.53 cm⁻¹ (C=C), 1403 cm⁻¹ (COO⁻), and 1045.64 cm⁻¹, whereas Bai showed bands at 3411.18 cm⁻¹ (OH), 3094.67 cm⁻¹ (C–H), 1659 cm⁻¹ (C=O), and 1620–1586 cm⁻¹ (C=C). The FTIR spectrum of Bai@GA nanofibers resembled that of pure GA, with a shifted O–H peak indicative of intramolecular hydrogen bonding, confirming self-assembly. Gel exhibited representative amide bands (3285, 1628, 1540, 1238 cm⁻¹), along with a COO⁻ peak at 1400 cm⁻¹. In Bai@GA/Gel/Zn²⁺, the broadened and shifted peak at 3267 cm⁻¹ suggested the formation of a hydrogen-bonded network between –NH₂, –OH, and –COOH groups of Gel and Bai@GA nanofibers, enhancing mechanical stability. Further supporting the structural evolution, the characteristic Bai@GA peaks at 1723 cm⁻¹ and 1657 cm⁻¹ disappeared in the Bai@GA/Gel/Zn²⁺ composite, replaced by a new band at 1628 cm⁻¹. This change confirms the coordination between Zn²⁺ and the carboxyl groups of GA, aligning with reported roles of multivalent metal ions in reinforcing hydrogel networks through complexation [32].

Rheological frequency sweeps (Fig. 6g) revealed that the storage modulus (G') consistently maintained higher values than the loss modulus (G'') throughout the investigated frequency range, validating the hydrogel’s robust mechanical properties. Strain sweep analysis (Fig. 6h) identified the linear viscoelastic regime. Based on comprehensive rheological characterization, the hydrogel containing 0.4 wt% Bai@GA was selected for all subsequent investigations due to its optimal mechanical performance. It exhibited the highest storage modulus and the greatest critical strain, indicating superior structural integrity and stability. The observed decline in storage modulus at the higher concentration of 0.6 wt% GA is presumably attributable to nanofiber overcrowding, which can introduce structural heterogeneity and compromise the gel network. Fig. S4a-c further manifested the hydrogel’s excellent viscoelasticity. Uniaxial compression tests revealed a Young's modulus of 33.086 kPa, with fracture occurring at 42% strain, indicating appropriate softness and deformability matching biological tissues. The adhesion performance was quantitatively evaluated via lap-shear testing, exhibiting a remarkable adhesive strength of 61.584 kPa on porcine skin, which ensures reliable attachment to wound surfaces. These results, combined with our rheological analyses demonstrating solid-dominated viscoelasticity (G' > G'') and excellent structural recovery, collectively validate the hydrogel's robust mechanical integrity and tissue-adhesive capability.

The Bai@GA/Gel/Zn²⁺ hydrogel exhibited clear pH-dependent swelling, degradation, and release behavior under physiologically relevant conditions. At pH 7.4, it reached a maximum swelling ratio of 186%, while under the acidic conditions (pH 5.5), swelling was markedly limited to 106% (Fig. S4d). This reduction correlates with the protonation of carboxyl groups in GA and Gel under acidic conditions, which diminishes electrostatic repulsion and enhances structural compactness, consistent with previously reported behavior of GA-based systems [32], [33]. To further assess the clinical relevance, these behaviors were investigated in a simulated wound exudate (SWS) containing 5% FBS, 10 ng/mL TNF-α, and 1 µg/mL MMP-9. In this complex milieu, the presence of serum proteins led to a increase in swelling (approximately 135%). Degradation was largely complete at pH 7.4 within the test period, whereas acidic conditions moderately delayed the process, though the hydrogel still degraded substantially within 72 h (Fig. S4e). It is noteworthy that the presence of proteases significantly accelerates the degradation of hydrogels. Drug release profiles confirmed this pH responsiveness: both Bai@GA and the composite hydrogel showed biphasic release, with the the incorporation of Gel and Zn²⁺ providing more sustained delivery (Fig. S4f-g). Notably, release was significantly slower at pH 5.5 than at pH 7.4, highlighting the system's ability to extend drug retention in an acidic wound microenvironment. Nevertheless, the release in SWS was faster than in PBS due to matrix erosion by MMP-9. These results collectively demonstrate that the hydrogel's performance is governed by a pH-modulated and enzyme-influenced kinetics. The material's ability to extend drug retention specifically in an acidic, protease-rich environment strongly supports its suitability for sustained therapy in the complex diabetic wound microenvironment.

Biocompatibility, biosafety, antibacterial efficacy, and ferroptosis Resistance of Bai@GA/Gel/Zn²⁺ nanofibrous hydrogel

To assess the cytocompatibility of the nanofibrous hydrogels, HUVECs treated with Bai, Bai@GA, GA/Gel/Zn²⁺, and Bai@GA/Gel/Zn²⁺ were analyzed using CCK-8 evaluation and live/dead cell staining. The findings demonstrated that Bai@GA/Gel/Zn²⁺ exhibited superior biocompatibility compared to the control group, significantly promoting cell proliferation (Fig. 6i, S5a). Additionally, hemolysis tests confirmed outstanding blood compatibility, with all test groups showing hemolysis rates below the 5% safety threshold (Fig. 6k). When applied to the wound surface, the Bai@GA/Gel/Zn²⁺ hydrogel undergoes progressive degradation through hydrolysis and enzymatic cleavage, thus released Bai is metabolized locally and systemically via glucuronidation and sulfation [34], with metabolites cleared renally. This localized action minimizes systemic exposure while maintaining therapeutic efficacy at the wound site. For biosafety assessment, histological evaluation of major organs (heart, liver, spleen, lungs, and kidneys) revealed normal tissue architecture and no observable pathological changes in all treatment groups (Fig. S5b). The robust antibacterial capacity of Bai@GA/Gel/Zn²⁺ ensured a sterile wound microenvironment. Typical examples of bacteria like S.aureus and E.coli were employed to validate this antibacterial activity. Antibacterial assessments revealed a progressively enhanced effect from Bai, GA/Gel/Zn²⁺, Bai@GA, to Bai@GA/Gel/Zn²⁺, demonstrating a stepwise improvement in inhibitory potency (Fig. 6j, l). While both Bai and GA/Gel/Zn²⁺ alone showed modest activity, their integration into the Bai@GA/Gel/Zn²⁺ composite yielded a remarkable bactericidal rate exceeding 70% against S.aureus, a performance superior to that observed against E.coli.

Furthermore, to evaluate the anti-ferroptosis potential of the hydrogels, HUVECs were treated with RSL-3 (a specific ferroptosis inducer) as the disease control and Fer-1 (a well-characterized inhibitor) as the positive control. As expected, Bai@GA/Gel/Zn²⁺ demonstrated the most potent protective activity among all groups, as evidenced by the most substantial reductions in ROS, MDA, and Fe²⁺ levels, nearly matching the efficacy of Fer-1 (Fig. 6m-p). By contrast, GA/Gel/Zn²⁺ alone showed only a marginal effect. These results clearly indicate that Bai is the primary active component against ferroptosis, and its efficacy is optimally preserved and delivered by the Bai@GA/Gel/Zn²⁺ composite system. To further delineate the specificity of Bai and Bai@GA/Gel/Zn²⁺ in regulating ferroptosis and to address potential crosstalk with other cell death mechanisms, additional assays were performed under H₂O₂-induced oxidative stress. Annexin V/PI flow cytometry and PI staining revealed that compared to the Ctrl group, treatment with Bai can significantly attenuated H₂O₂-induced apoptosis and secondary necrosis. This protective effect was substantially enhanced in the Bai@GA/Gel/Zn²⁺ group (Fig. S6a-d), indicating the advantage of the composite hydrogel. Mechanistically, WB analysis revealed that both Bai and Bai@GA/Gel/Zn²⁺ treatments markedly activated the PI3K/Akt survival pathway, evidenced by significantly increased ratios of p-PI3K/PI3K and p-Akt/Akt compared to the injured control (Fig. S6e-g). This activation correlates with the observed reduction in apoptosis, suggesting a contributory role of this pathway to the overall cytoprotection. In summary, the cytoprotection afforded by Bai and Bai@GA/Gel/Zn²⁺ is multifaceted. The defining mechanism is the specific reversal of ferroptotic hallmarks, with PI3K/Akt-mediated attenuation of apoptosis acting as a complementary pathway.

Current research has developed various hydrogels based on natural polymers for the delivery of Bai. Nie et al. developed a rapidly in situ-gelling Schiff base hydrogel, albeit with limited mechanical strength [35]; Qin et al. created a ZIF-8-based nanocomposite hydrogel that actively restores mitochondrial function and promotes angiogenesis [36]; while Shang et al. developed a pH-sensitive hydrogel from chitosan that combats infection and inhibits scarring by repolarizing macrophages [37]. The Bai@GA/Gel/Zn²⁺ hydrogel developed in our study features a dual physically cross-linked network formed by Zn²⁺ metal coordination bonds and hydrogen bonds, which provides exceptional mechanical strength and viscoelasticity. It enables sustained release of Bai in the acidic wound microenvironment, comprehensively exerting antibacterial, antioxidant, and anti-ferroptosis effects, thereby demonstrating significant potential for managing complex diabetic wounds.

Bai@GA/Gel/Zn²⁺ nanofiber hydrogel accelerates diabetic wound healing in vivo

Based on the aforementioned findings, to evaluate the therapeutic potential of Bai@GA/Gel/Zn²⁺ in vivo, a diabetic rat wound model was established. Surgical excision was performed to create round cutaneous wounds (diameter: ∼10 mm) penetrating all skin layers on the dorsal region of diabetic rats (Fig. 7a) followed by therapeutic administration of PBS, Bai, Bai@GA, GA/Gel/Zn²⁺ or Bai@GA/Gel/Zn²⁺. Relative to the untreated group, all intervention groups (Bai, Bai@GA, GA/Gel/Zn²⁺ or Bai@GA/Gel/Zn²⁺) displayed accelerated wound healing (Fig. 7b). Detaily, the Bai@GA/Gel/Zn²⁺ group achieved the fastest healing rate, with a wound closure of 96.4% by day 9, significantly surpassing GA/Gel/Zn²⁺ (89.1%), Bai@GA (84.6%) and Bai (76.5%) groups (Fig. 7e). Histological analysis via HE staining revealed markedly reduced wound width and complete epithelialization in the Bai@GA/Gel/Zn²⁺-treated group. (Fig. 7c, f). Masson staining was employed to evaluate collagen fiber deposition and alignment. The results revealed abundant collagen deposition with well-organized architecture in the Bai@GA/Gel/Zn²⁺ group, demonstrating superior performance compared to other groups (Fig. 7d, g).

Fig. 7.

Bai@GA/Gel/Zn²⁺ Nanofiber Hydrogel Accelerates Diabetic Wound Healing In Vivo. (a) Schematic illustration of the wound model in diabetic rats. (b) Representative images of wound healing. (c-d) Images of HE and Masson staining. Scale bar was 1 mm and 100  μm. (e) Wound area quantification. n = 5. (f-g) Analysis of tissue width in HE staining and collagen deposition in Masson staining. n = 5. Data was analysed using one way ANOVA or two way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

Dual IF staining was used to further analyze the vascularization status in wound tissues. It was observed that the Bai@GA/Gel/Zn²⁺ group exhibited greatly upregulated expression of CD31/α-SMA co-expression and increased neovascularization(Fig. 8a, f). Concurrently, decreased TNF-α levels in this group (Fig. S8a-b) suggested anti-inflammatory activity, that synergistically promoted wound healing. To investigate the molecular mechanisms, we analyzed key ferroptosis markersand the Keap1/NRF2/HIF-1α pathway in wound tissues via WB and IHC (Fig. 8b-e, 8g-i, S7a-b, S8c). Consistent with our in vitro findings, Bai@GA/Gel/Zn²⁺ treatment significantly inhibited ferroptosis, as evidenced by upregulated GPX4 and downregulated NCOA4 and TFR1. Mechanistically, these effects were mediated through the Keap1/NRF2/HIF-1α axis, wherein Bai@GA/Gel/Zn²⁺ markedly reduced Keap1 levels while enhancing the expression of both NRF2 and HIF-1α. The composite system exerted the most profound anti-ferroptotic activity among all treatment groups, closely mirroring the efficacy trend observed in cellular assays. By contrast, GA/Gel/Zn²⁺ alone only marginally modulated these targets. These data collectively underscore that Bai serves as the principal anti-ferroptotic component, and its bioactivity is effectively preserved and enhanced within the Bai@GA/Gel/Zn²⁺ delivery system, leading to robust activation of cytoprotective signaling in the wound microenvironment.

Fig. 8.

Mechanism of Bai@GA/Gel/Zn²⁺ in promoting diabetic wound healing. (a) Dual IF images of CD31 and α-SMA. Scale bar: 100  μm. (b-e) IHC images of GPX4, NCOA4, TFR1, and HIF-1α. Scale bar: 100  μm. (f) Quantitative statistics of blood vessels. n = 3. (g-i) Quantitative expression analysis of GPX4, NCOA4, TFR1, and HIF-1α. n = 3. Data was analysed using one way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

In summary, the Bai@GA/Gel/Zn²⁺ nanofiber hydrogel enables efficient delivery and sustained release of Bai, promoting angiogenesis, suppressing ferroptosis, and accelerating diabetic wound healing in vivo through a mechanism potentially linked to Keap1/NRF2/HIF-1α signaling. These findings highlight its translational potential as a multifunctional hydrogel for chronic wound.

Conclution

This study elucidates that Bai suppresses endothelial ferroptosis through modulating the Keap1/NRF2/HIF-1α signaling axis, thereby restoring angiogenesis by preserving endothelial cell viability, migration and redox homeostasis, ultimately accelerating wound closure in diabetic models. To circumvent its low bioavailability, we developed a GA nanofiber-reinforced hydrogel for controlled drug delivery, enhancing cellular proliferation, antibacterial activity, antioxidant defense, and ferroptosis inhibition. These findings provide a multifaceted therapeutic strategy targeting the pathological interplay between ferroptosis and impaired vascularization in diabetic complications.

Data availability statement

Data from this research can be accessed by contacting the corresponding authors, subject to validation of request rationale.

Ethic approval and consent to participate

The Institutional Animal Use and Care Committee of the Nanjing Durm Tower Hospital consented to all experiments on animals (Ethics Approval Number: 2025AE02005).

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.

Appendix A. Supplementary data

The following are the Supplementary data to this article: