Multifunctional wound dressing for highly efficient treatment of chronic diabetic wounds

Abstract

As the number of diabetics increases yearly and diabetic ulcer is the most concerned diabetic complication, it is of great urgency to develop a novel multifunctional material with high efficiency to promote diabetic wound healing. In this paper, ceria (CeO₂)‐modified upconversion nanoparticles which loaded in polycaprolactone membranes were successfully prepared as a ceria‐based wound dressing for chronic wound healing. In this system, near‐infrared (NIR) light is used as a switch to control the photodynamic antibacterial therapy by converting the NIR photons into ultraviolet light, which can further trigger CeO₂ to produce reactive oxygen species. In addition, the well‐known antioxidant ability of CeO₂ endows the wound dressing with prominent capability to resist the oxidative stress around chronic wounds. In summary, both photodynamic antibacterial therapy and good antioxidant property are realized simultaneously in this CeO₂‐based system to promote the healing of diabetic chronic wounds.

INTRODUCTION

Wound healing is a complex dynamic process consisting of several overlapping stages, including hemostasis, inflammation, proliferation (granulation stage), and maturity (remodeling stage).^1–3 The inflammatory stage of diabetic chronic wounds is prone to be prolonged by the surrounding bacteria growth, which inhibits the normal process of wound healing. Traditional antibiotic therapy is the most widely used method for bacterial infected wound, but it can induce the formation of drug resistant bacteria and further increase the difficulty of later treatment.^4–6 Photodynamic therapy (PDT), dependent on reactive oxygen species (ROS), is a newly developed method to deal with bacteria and even biological film and holds high efficiency without side effects.^7–10 Although moderate ROS is beneficial to normal wound healing by alleviating inflammation and stimulating cell migration and angiogenesis, excess ROS will inhibit the normal wound healing process and lead to the formation of chronic wounds.^11,12 Diabetic wounds are a kind of chronic wounds, which exhibit sustained inflammatory response and excess accumulation of ROS. The special microenvironment around the wounds further prevents the transition of healing from the inflammatory stage to proliferation stage.¹³ Therefore, adding nanomaterials with antioxidant function to wound dressing is a strategy to accelerate chronic wound repair. The dressing with antioxidant property can effectively remove excess ROS around chronic wounds and maintain normal cell functions.¹⁴

Cerium (Ce) is the most abundant rare earth element, and ceria (CeO₂) is one of the most important and representative cerium compounds with the fluorite structure, which has been widely used in the fields of catalysis, sensor, biomedicine, and cosmetics.^15-19 The reversible conversion between Ce³⁺ and Ce⁴⁺ endows CeO₂ with high oxygen storage capacity and enzymatic activity, showing antioxidant ability to clear away excess ROS around chronic wounds.^20,21 Yan's group reported that the chitosan-coated CeO₂ nanocubes can promote healing of diabetic wounds by decreasing the persistence time of the inflammatory state and increasing the level of antioxidant enzymes.²² Besides, CeO₂ also exhibits the inherent antibacterial property by destroying the cell membranes of bacteria, but it is too low to resist bacteria around the diabetic chronic wounds.^23-25 Researchers have found that CeO₂ can be stimulated by ultraviolet (UV) light to produce ROS, which has been widely used in catalysis, but the side-effects of directly using UV light as light source to biological tissues limit the further application of this characteristic in biomedical engineering.^26-28

Upconversion nanoparticles (UCNPs) are a type of luminescent nanostructures, which can convert two or more low-energy continuous near-infrared (NIR) photons into short wavelength photons with higher energy effectively.^29,30 It means that UCNPs can efficiently convert NIR light with lower energy and stronger penetrating ability to UV or visible (vis) light, avoiding the direct damage of using UV light to biological tissues.³¹ Generally, UCNPs are compounds doped with different rare earth ions (Ho³⁺, Er³⁺, Tm³⁺, etc.) owning some special properties, such as imaging and magnetic resonance.^32-34 Combining good biocompatibility, the advantages listed above provide UCNPs with the superiority for biomedical applications. It is interesting that lithium-based UCNPs not only exhibit attractive upconversion luminous efficiency, but also have stronger emission intensity in the UV light-wave bands.^35-38 Meanwhile, the UV light produced by lithium-based UCNPs under NIR irradiation can be used to trigger CeO₂ to yield adequate ROS for PDT to resist bacteria.^39,40

Herein, we report a novel lithium-based UCNPs (UCNPs@SiO₂@CeO₂) for wound healing (Scheme 1A). Under 980 nm NIR irradiation, UCNPs can be stimulated to produce UV light at about 350 nm, which can further trigger CeO₂ to produce adequate ROS to realize PDT. In the inflammatory stage of diabetic chronic wounds healing, UCNPs@SiO₂@CeO₂ loaded polycaprolactone (PCL) membranes (UCNPs@SiO₂@CeO₂/PCL membranes) exhibit excellent ability to resist bacteria under 980 nm NIR irradiation. As time goes on, excessive inflammatory response around diabetic chronic wounds continually inhibits wound healing. The antioxidant performance of CeO₂ endows it with the ability of removing ROS and making the transformation of wounds from the inflammatory stage to the proliferation stage to promote wound healing. Interestingly, PDT to against bacteria and antioxidant ability are realized simultaneously in this CeO₂-based system, and 980 nm NIR laser can be used as a “switch” for the efficient photodynamic antibacterial ability. The schematic diagram of the mechanism is shown in Scheme (1B).

EXPERIMENTAL SECTION Materials

Trifluoroacetic acid (CF₃COOH, 99%) and thiazolyl blue (98%) were purchased from RHAWN, Shanghai, China. LB nutrient agar, tryptone soy broth (TSB), rhodamine-labeled phalloidin and Masson's trichrome stain kit were obtained from Solarbio Science & Technology Co., Ltd, Beijing, China. 1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (C₈H₁₇N₃·HCl, EDC, 99%) and N-hydroxysuccinimide (C₄H₅NO₃, NHS, 98%) were purchased from 3A Chemicals, Shanghai, China. 4-Morpholinoethanesulfonic acid (C₆H₁₃NO₄S, MES, 99%) and sodium acetate (CH₃COONa·3H₂O, 99%) were purchased from InnoChem Science & Technology Co., Ltd, Beijing, China. PCL ((C₆H₁₀O₂)_n, PCL, M_w = 80,000) was purchased from Shanghai Yuanye Bio-Technology Co., Ltd, Shanghai, China. Ceric ammonium nitrate ((NH₄)₂Ce(NO₃)₆, 99.99%), paraformaldehyde ((CH₂O)_n, AR), ytterbium oxide (Yb₂O₃, 99.9%), yttrium oxide (Y₂O₃, 99.99%), thulium oxide (Tm₂O₃, 99.9%), diethylene glycol (C₄H₁₀O₃, 98%), (3-aminopropyl)triethoxysilane (H₂NCH₂CH₂Si(OC₂H₅)₃, APTES, 98%), Igepal CO-520 ((C₂H₄O)_n·C₁₅H₂₄O, n ∼ 5), ammonia solution (NH₄OH, 28.0%–30.0% NH₃ basis), poly(acrylic acid) ((C₃H₄O₂)_n, PAA, M_w∼3,000, 50% solution), isopropyl alcohol (C₃H₈O, AR, ≥99.5%), and cyclohexane (C₆H₁₂, AR, ≥99.5%) were purchased from Aladdin Bio-Chem Technology Co., Ltd, Shanghai, China. Tetraethyl orthosilicate (C₈H₂₀O₄Si, TEOS, AR) was bought from Sinopharm Chemical Reagent Co., Ltd, Shanghai, China. N, N-Dimethyl formamide (C₃H₇NO, DMF, AR) was bought from Tianjin Damao Chemical Reagent Factory, Tianjin, China. Acetic acid (CH₃COOH, AR, ≥99.5%), dimethyl sulfoxide (C₂H₆OS, DMSO, GR, 99.8%), n-hexane (C₆H₁₄, AR, ≥97.0%), and ethanol (C₂H₅OH, AR) were purchased from Tianjin Concord Technology Co., Ltd, Tianjin, China. Chloral hydrate (CCl₃CH(OH)₂, AR, >99.0%) was purchased from Shanghai Macklin Biochemical Co., Ltd, Shanghai, China. Toluene (C₇H₈, AR, 99.5%) and trichloromethane (CHCl₃, AR, ≥99.0%) were obtained from No. 6 Tianjin Chemical Reagent Factory, Tianjin, China. Lithium trifluoroacetate (CF₃COOLi, 97%) was bought from Alfa Aesar Chemical Co., Ltd, Shanghai, China. Oleic acid (C₁₈H₃₄O₂, OA, 90%), oleylamine (C₁₈H₃₇N, OM, 70%) and 1-octadecene (C₁₈H₃₆, ODE, 90%) were obtained from Sigma–Aldrich Trading Co., Ltd, Shanghai, China. Cell experiments related reagents, including cell culture medium Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), phosphate buffered saline (PBS), trypsin, penicillin, and streptomycin were purchased from Gibco Company (Grand Island, NY, USA). Hematoxylin and eosin staining kit and 4',6-diamidino-2-phenylindole dihydrochloride (DAPI) were obtained from Beyotime Biotechnology Co., Ltd, Shanghai, China. Calcein-AM/propidium iodide (PI) was purchased from DOJINDO Laboratories, Japan. All chemicals were used as received. The water used in experiments was purified by a GenPure UV/UF water purification system (Thermo Scientific, Germany). Staphylococcus aureus and Escherichia coli were obtained from the School of Chemical Engineering and Technology, Tianjin University (Tianjin, China), and stored at −80°C until used.

Preparation of composite membranes Preparation of ceria (CeO₂)

CeO₂ was synthesized according to the reported literature with a slight modification.⁴¹ In a typical procedure, 2.19 g of Ce(NH₄)₂(NO₃)₆ and 8 g of CH₃COONa were dissolved in 56 ml of water in a beaker with stirring. Then, 8 ml of CH₃COOH was dissolved in the above mixed solution followed with continuous stirring at room temperature for 30 min and the mixture was transferred to a Teflon lined autoclave for hydrothermal reaction at 220°C for 12 h. After reaction and cooling to room temperature, pale yellow precipitate was collected by centrifugation (8000 rpm, 10 min) and washed with water and ethanol for several times to obtain the pure products. Finally, the obtained products were well dispersed in water by ultrasound and then freeze-dried to get CeO₂ powders.

Preparation of carboxyl modified CeO₂ (CeO₂–COOH)

Carboxyl modified CeO₂ (CeO₂–COOH) was prepared by ligand exchange reaction between PAA and CeO₂.⁴² First, 600 mg of PAA (50% solution) and 30 ml of diethylene glycol were added into a three-necked flask (50 ml). Ar was filled into the three-necked flask when it was heated to 110°C under the conditions of magnetic stirring and vacuum state. Then, 2 ml of toluene solution dispersed with CeO₂ (50 mg ml^–1) was slowly dropped into the three-necked flask, and the reaction was maintained at 110°C for 60 min. After the reaction proceeded at 240°C for 90 min, the system was cooled to room temperature naturally. Precipitate can be obtained after ethanol was added into the system, which was collected by centrifugation, washed with water three times and dried at 60°C to obtain pure CeO₂–COOH.

Preparation of precursor (rare earth trifluoroacetate (RE(CF₃COO)₃))

The synthesis of UCNPs was adjusted based on the methods reported in the literature.⁴³ The first step was to prepare the precursor—rare earth trifluoroacetate (RE(CF₃COO)₃), which was prepared by refluxing rare earth oxides (RE₂O₃, RE = Y, Tm, Yb) in excess CF₃COOH solution (the volume ratio between CF₃COOH and water was 1:1) at 80°C. After the white suspension was thoroughly clarified, the system was cooled down to room temperature and filtered. Then, the filtrate was put into the oven and dried at 120°C to obtain white powder of RE(CF₃COO)₃, which was collected and used later.

Synthesis of LiYbF₄:Tm³⁺@LiYF₄

LiYbF₄:Tm³⁺ was synthesized via a previously reported thermal decomposition method with minor modification [36]. 1 mmol RE(CF₃COO)₃ (99.5% Yb(CF₃COO)₃ and 0.5% Tm(CF₃COO)₃) and 1 mmol CF₃COOLi were mixed with 5 ml OA and 5 ml ODE in a 50 ml three-necked flask at room temperature. Under vacuum and magnetic stirring, the mixed solution was heated to 120°C at a rate of 10°C min^–1 by a program-controlled heating device and kept for 30 min to remove water and oxygen until a transparent solution formed. After the flask was filled with Ar, the reaction system was heated to 320°C at a rate of 5°C min^–1 and kept for 1 h. After reaction, the system was cooled to room temperature under Ar protection and 15 ml ethanol was added. The obtained precipitate was collected by centrifugation and washed with CHCl₃/ethanol twice, and the product was dried in a vacuum drying oven to get LiYbF₄:Tm³⁺. The hot-injection method was used to prepare LiYbF₄:Tm³⁺@LiYF₄. Solution A was prepared in a 50 ml flask by dissolving 1 mmol RE(CF₃COO)₃ (99.5% Yb(CF₃COO)₃ and 0.5% Tm(CF₃COO)₃) and 1 mmol CF₃COOLi in 10 ml OA/ODE solution (the volume ratio between OA and ODE was 1:1). Solution B was prepared by mixing 1 mmol of Y(CF₃COO)₃ and 1 mmol of CF₃COOLi in 5 ml OA/ODE solution (the volume ratio between OA and ODE was 1:1). Both solution A and solution B were stirred and heated to 120°C in vacuum for 30 min. The temperature of solution A was raised to 320°C under Ar atmosphere and kept for 1 h and then solution B was injected via a pump–syringe system at a rate of 0.5 ml min^–1 and kept for another 1 h. After cooling, washing, and drying, LiYbF₄:Tm³⁺@LiYF₄ nanoparticles can be obtained which are labeled as UCNPs in this paper.

Preparation of silica (SiO₂) coated UCNPs (UCNPs@SiO₂)

Two hundred milligram UCNPs were dispersed in 60 ml cyclohexane containing 1 ml Igepal CO-520, and stirred for 10 min. Then 4 ml Igepal CO-520 and 0.8 ml NH₄OH were added, and the substances were mixed uniformly by ultrasound for 20 min. After the solution became clear, 0.4 ml TEOS was added dropwise, and stirred at room temperature for 48 h. The product was collected by centrifugation, washed twice with H₂O/C₂H₅OH (v/v 1/1) mixed solution, and dried in an oven at 60°C to obtain silica (SiO₂) coated UCNPs, which are labeled as UCNPs@SiO₂ in this paper.

Preparation of amino modified UCNPs@SiO₂ (UCNPs@SiO₂-NH₂)

The amino functionalization of UCNPs@SiO₂ was achieved by hydrolysis of adding APTES during the reflux reaction. First, 200 mg of the as-prepared UCNPs@SiO₂ were added into a 250 ml three-necked flask, and 200 ml isopropanol was added followed with ultrasonic dispersion. After the system was heated to 85°C, 1 ml APTES was added dropwise followed with reflux. The system was cooled to room temperature naturally after 6 h reaction, and the amino functionalized UCNPs@SiO₂ were collected after centrifugation, washing, and freeze-drying, which are labeled as UCNPs@SiO₂-NH₂ in this paper.

Preparation of CeO₂-modified UCNPs@SiO₂ (UCNPs@SiO₂@CeO₂)

CeO₂-modified UCNPs@SiO₂ can be prepared by amidation reaction between UCNPs@SiO₂-NH₂ and CeO₂-COOH. First, 25 mg of CeO₂–COOH was dispersed in 20 ml of MES buffer (10 mM, pH = 5.5) including 10 mM EDC and 25 mM NHS, and incubated at 37°C for 15 min. Subsequently, a certain amount of UCNPs@SiO₂-NH₂ was added into the above solution and stirred at 37°C for 12 h. Finally, the products were washed with water for three times after centrifugation and dried at 60°C to obtain CeO₂-modified UCNPs@SiO₂, which are labeled as UCNPs@SiO₂@CeO₂ in this paper. The mass ratios between UCNPs@SiO₂-NH₂ and CeO₂-COOH are controlled to be 2:1, 1:1, 1:2, and 1:3 during reaction, which are labeled as UCNPs@SiO₂@CeO₂ (2-1), UCNPs@SiO₂@CeO₂ (1-1), UCNPs@SiO₂@CeO₂ (1-2), and UCNPs@SiO₂@CeO₂ (1-3), respectively.

Preparation of UCNPs@SiO₂@CeO₂/PCL membranes

UCNPs@SiO₂@CeO₂ containing PCL solution was used to prepare membranes by the electrospinning method. First, PCL was dissolved in DMF/CHCl₃ solution (volume ratio between DMF and CHCl₃ was 1:9) with 10% (w/v) concentration. Then, UCNPs@SiO₂@CeO₂ (UCNPs or CeO₂) were dispersed in the above solution (weight ratio between nanoparticles and PCL was 1:9) by ultrasonic for electrospinning. For the electrospinning device, the aluminum foil was connected with the negative electrode as the receiving end, and the syringe needle was connected with the positive electrode as the transmitting end. The electrostatic voltage was 15 kV, the distance between the receiving plate and the needle was 15 cm, and the propulsion speed of the peristaltic pump was 0.02 ml min^–1. After spinning, the aluminum foil at the receiving end was placed in a vacuum oven at 40°C for 3 h to remove the residual organic solvent, and then the membranes can be collected. Depending on the difference in composition, the membranes are marked as UCNPs/PCL, CeO₂/PCL, and UCNPs@SiO₂@CeO₂/PCL, respectively.

Basic characterizations

The morphologies of the nanoparticles were detected using a JEOL JEM-2800 (Japan) low- and high-resolution transmission electron microscope (TEM and HRTEM) operating at 200 kV. The LiYbF₄:Tm³⁺ and LiYbF₄:Tm³⁺@LiYF₄ were dispersed in CHCl₃ and dropped onto a copper grid for TEM tests, while other samples were dispersed in ethanol. The morphologies of the membranes and dead bacteria were examined using field emission scanning electron microscope (FESEM, JEOL JSM-7800F, Japan) with an acceleration voltage of 5 kV. The crystalline phases of the nanoparticles were determined via X-ray diffractometer (XRD) analysis on Rigaku Smart Lab 3 kW (Japan) using Cu-Kα radiation, and the scan area was from 10° to 80° with a speed of 5° min^–1. The functional groups on the surface of nanoparticles were recorded by Fourier transform infrared (FT-IR, Brucker-Tensor 37, Germany). Upconversion emission spectra of 1 mg ml^–1 hexane dispersions of UCNPs, UCNPs@SiO₂, and UCNPs@SiO₂@CeO₂ were obtained at room temperature under 980 nm NIR laser (Wave Particle Optoelectronic Technology Co., Ltd, Beijing, China) irradiation by a FL970 Fluorescence Spectrometer (Techcomp, China). The UV–vis absorption spectra were characterized on UV–vis spectrophotometer (SHIMADZU, UV-2600i, Japan). X-ray photoelectron spectroscopy (XPS) measurement was obtained using an ESCALAB 250Xi photoelectron spectrometer with a standard Al anode (Thermo Scientific, USA).

Stability of composite membranes

The stability of UCNPs@SiO₂@CeO₂/PCL fiber membranes was tested to investigate the durability of nanoparticles in PCL fibers. The fiber membranes of different compositions were soaked in PBS and incubated at 37°C in a shaker, and photographs were taken on Days 0, 2, and 4, respectively.

Antibacterial activity evaluation in vitro

The antibacterial activity of nanoparticles against Gram-negative bacteria (E. coli) and Gram-positive bacteria (S. aureus) under different irradiation intensity was first tested by the spread plate method. All the nanoparticles were sterilized before use, and the antibacterial experiments were done as follows. The nanoparticles with a final concentration of 2 mg ml^–1 and diluted bacteria suspension (4.0 × 10⁵ CFU ml^–1) were added into the 96-well plates including PBS, which were further placed in a shaker at 37°C and 160 rpm for 10 min for the complete contact between bacteria and materials. After that, the 96-well plates were irradiated under 980 nm NIR laser for antibacterial test in a biological safety cabinet. The different irradiation times of 0, 5, 10, and 15 min at each optical power density and the optical power density of 0.8, 1.0, and 1.2 W cm^–2 were chosen as the parameters to derive the optimal irradiation intensity. Based on the antibacterial results of nanoparticles, the antibacterial activity of PCL, UCNPs/PCL, CeO₂/PCL, and UCNPs@SiO₂@CeO₂ (2-1, 1-1, 1-2, 1-3)/PCL against E. coli and S. aureus were also measured by the spread plate method. Each well was irradiated for different times with an optical power density of 1.0 W cm^–2. After the irradiation, the bacterial solutions were diluted by 1.0 × 10⁴ times and 80 µl of the suspensions were further used in the spread plate method.

More intuitively, bacteria morphologies before and after the treatment of 980 nm NIR irradiation were further observed by FESEM, and UCNPs@SiO₂@CeO₂ (1-1)/PCL group was used as the representative of the UCNPs@SiO₂@CeO₂/PCL groups. After antibacterial experiments, the membranes were washed with PBS twice, and fixed with 4% paraformaldehyde at room temperature overnight. And then, all the membranes were washed with PBS twice, and dehydrated with 10%, 20%, 30%, 50%, 70%, 90%, and 100% (v/v) ethanol for 30 min respectively in gradient. Finally, the dehydrated membranes were adhered to the conductive tape for FESEM observation after gold coating.

Cell culture

The cells involved in this experiment were L929 cell lines (mouse fibroblast cell lines) which were provided by School of Materials Science and Engineering, Tianjin University. Cells were cultured in the DMEM medium supplemented with 10% FBS, 100 µg ml^–1 streptomycin and 100 U ml^–1 penicillin at 37°C in a humidified incubator with 5% CO₂.

Detection of reactive oxygen species

The capacities of CeO₂, UCNPs, and UCNPs@SiO₂@CeO₂ to produce ROS in vitro were detected using 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) assay. For qualitative measurement, 0.25 ml of DCFH-DA solution (1.0 mM, dissolved in ethanol solution) was added into 1.0 ml NaOH solution (0.01 mM, dissolved in ethanol solution), and stirred at room temperature for 30 min to convert DCFH-DA into 2′,7′- DCFH. After neutralizing the above hydrolysate with 5 ml PBS solution, the as-prepared solution was stored at −20°C for later use. During the experiment, 40 µl of the as-prepared DCFH solution was added into 3 ml of nanoparticle aqueous solution (2 mg ml^–1) to form a mixed solution, which followed by stirring in dark for 30 min to achieve the absorption–desorption equilibrium. Subsequently, the solution was irradiated with 980 nm laser (1.2 W cm^–2) and the fluorescence emission spectra were recorded by a FL970 fluorescence spectrometer at 0 min, 10 min, 20 min, and 30 min, respectively. The fluorescence spectrum is measured in the range of 510–650 nm under 488 nm excitation, and the maximum emission peak appears at near 525 nm. As more DCFH was oxidized by ROS into 2′,7′-dichlorofluorescein (DCF), much higher fluorescence intensity can be detected.

In order to confirm the ROS scavenging capability, DCFH-DA assay was done in cells. After entering cells, DCFH-DA can be hydrolyzed by cellular esterase into DCFH with no fluorescence which can be further oxidized by ROS to generate fluorescent DCF. The UCNPs, CeO₂, and UCNPs@SiO₂@CeO₂ were added to each well in a 24-well plate containing 5 × 10⁵ cells per well with a final concentration of 2 mg ml^–1 and incubated at 37°C for 24 h in an incubator. Then, 0.5 ml of H₂O₂ containing DMEM solution (200 µM) was added to each well. After incubating at 37°C for 120 min and washed with PBS twice, 250 µl of DCFH-DA with final concentration of 10 µM (in FBS-free DMEM) was added and incubated for another 20 min. Each well was washed with PBS thrice to remove the unreacted DCFH-DA. Cells cultured without H₂O₂ solution and with H₂O₂ solution only were set as the control groups. The fluorescence signals of the cells were observed by inverted fluorescence microscope (LEICA DMi8, Germany).

Cytotoxicity assay

Thiazolyl blue tetrazolium bromide (MTT) assay was performed to investigate the cytotoxicity of PCL, UCNPs/PCL, CeO₂/PCL, and UCNPs@SiO₂@CeO₂ (2-1, 1-1, 1-2, 1-3)/PCL. All the membranes were sterilized by 75% alcohol (v/v) and washed with sterile PBS several times in the 96-well culture plates before used. L929 cells were seeded in 96-well plates with a density of 1.0 × 10⁴ cells per well and incubated at 37°C for 24 h under 5% CO₂. After removing medium, cells were treated with MTT for another 4 h (final concentration was 0.5 mg ml^–1), and then DMSO was added to dissolve the formed formazan crystals. Finally, microplate reader was used to measure the absorbance at 490 nm. Three groups of each sample were tested as parallel experimental data.

In order to further investigate the viability of L929 cells after cultured on PCL, UCNPs/PCL, CeO₂/PCL, and UCNPs@SiO₂@CeO₂/PCL membranes, the live-dead assay was implemented. L929 cells were seeded on the sterile membranes and incubated at 37°C for 24 h, and were further treated with Calcein-AM and PI for observation by the inverted fluorescence microscope. The excitation filter was set at 490 nm to observe living (green) cells labeled by Calcein-AM and at 535 nm to observe dead (red) cells labeled by PI.

Cell morphology

The cytoskeleton assay was applied to further investigate the modality of L929 cells after cultured on sterilized membranes. Cells seeded on the membranes were dyed with rhodamine-labeled phalloidin and DAPI respectively for cytoskeleton and cell nucleus labeling respectively after fixation with 4% paraformaldehyde and permeation by 0.1% Triton. Subsequently, the cells were observed by the inverted fluorescence microscope.

In addition, the morphology of the L929 cells seeded on the membranes was further observed by FESEM. The membranes seeded with L929 cells were fixed with 4% paraformaldehyde overnight, and the subsequent dehydration procedure was the same as the treatment of the bacteria for FESEM characterization.

Diabetic skin wound healing assay in vivo

Diabetic rat models with wounds on the back were used to investigate the promoting efficiency of the as-prepared membranes to diabetic wounds. Twenty male Sprague Dawley (SD) rats (200–300 g) were obtained from Vital River Laboratory Animal Technology Co., Ltd, Beijing, China. All the procedures were approved by the Laboratory Animal Center of Nankai University (Approval No. SYXK (Jin) 2019-0001). The type I diabetes model rats were constructed by injecting streptozotocin (STZ, ≥98%%, Reagent grade, Beyotime Biotechnology Co., Ltd, Shanghai, China) through the intraperitoneal injection (40 mg kg^–1 day^–1) for about 3 days until the blood glucose level was above 16.7 mM. The successfully modeled rats were randomly divided into two groups—non-irradiation group and 980 nm NIR irradiation group. After anesthetization and shaving, four circular wounds with 16 mm in diameter were created on the back of each rat. The four wounds were treated with control (without any treatment), CeO₂/PCL, UCNPs/PCL, and UCNPs@SiO₂@CeO₂ (1-1)/PCL, respectively and were fixed with elastic self-adhesive bandage. Different from the non-irradiation group, wounds in irradiation group were irradiated by 980 nm NIR for 15 min (1.0 W cm^–2) before bandaging. The day of the operation was recorded as day 0. On days 3, 5, 8, 10, and 15, the wounds were photographed, and the healing rate and remaining area of the wounds were calculated by Image J. In addition, on the days 3, 8, 10, and 15, tissues at the edge of wounds were collected and fixed in 4% paraformaldehyde solution for subsequent histological analysis.

In vivo infected wound healing

To assess the effects of UCNPs@SiO₂@CeO₂ (1-1)/PCL with irradiation of 980 nm NIR light in the infected wound, full-thickness cutaneous defects on the dorsal of male SD rats which were infected with S. aureus were established. The circular wound with a diameter of 10 mm was created and 100 µl S. aureus (∼1 × 10⁸ CFU ml^–1) suspension was inoculated onto the wound area. After 2 days, infected wounds were obtained and was marked as day 0. The infected wounds without any treatment, treated with NIR, treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL, and treated with both UCNPs@SiO₂@CeO₂ (1-1)/PCL and 980 nm NIR irradiation were marked as control, control+NIR, 1-1/PCL, and 1-1/PCL+NIR, respectively. In additional, wounds in the NIR group were irradiated by 980 nm NIR for 15 min (1.0 W cm^–2) before bandaging. The wounds were photographed on days 0, 3, 5, 7, 10 for further analysis. A little tissue on the surface of wound was taken down and immersed in 500 µl PBS buffer solution for 10 h, and the 80 µl extracts were spread on solid LB agar plate after diluting two times. After incubating at 37°C for 24 h, the agar plates were photographed.

Histological evaluation

After fixation, samples were embedded in paraffin and cut into 5 µm sections (KD-BM Tissue Embedding, and KD-2258 Rotary Microtome, Zhejiang Jinhua Kedee Instrumental Equipment Co. Ltd). Hematoxylin-eosin (H&E) staining and Masson's trichrome staining were performed to detect the growth of new tissues. Images were taken using a microscope (OLYMPUS CK31, Japan), and CaseViewer was used for image processing.

Statistical analysis

All the experiments were performed in triplicate. The obtained data were expressed as the mean value ± standard deviation, the statistical significance for multiple groups was analyzed by the one-way ANOVA followed by the Tukey’ test, and significance was achieved using Origin 2017 (*p <.05, **p <.01, and ***p <.001).

Data availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.

RESULTS AND DISCUSSION Basic characterization of nanoparticles

CeO₂ nanoparticles were successfully prepared by the hydrothermal method and they are cubic with the particle size of about 5 nm (Figure S1A). The interplanar spacing of 0.32 nm and 0.27 nm corresponds to (1 1 1) and (2 0 0) crystal planes, respectively, which can be seen from the HRTEM image (Figure S1B). In addition, the morphologies of carboxyl modified CeO₂ (CeO₂–COOH) nanoparticles have no change after carboxylation of CeO₂ (Figure S1C). The difference of diameter of particles is shown in Figure (S2). The morphologies of the nanoparticles before and after hot-injection are shown in Figure (S2A,C). The nanoparticles are rhombus and the nanoparticle size of UCNPs increases from 41.4 ± 2.1 nm (of LiYbF₄: Tm³⁺) to 68.7 ± 3.9 nm after the hot-injection (Figure S2B,D). Compared with UCNPs (Figure S2C), a transparent outer shell can be observed around UCNPs after the coating of UCNPs with silica (UCNPs@SiO₂) (Figure S3A), but the characteristic peak positions in the XRD pattern do not change or shift (Figure 1A). Aminated UCNPs@SiO₂ nanoparticles (UCNPs@SiO₂-NH₂) are presented in Figure (S3B) and amination does not change the morphology of nanoparticles. In the Fourier transform infrared (FTIR) spectrum (Figure 1B) of UCNPs@SiO₂, the characteristic peaks at 1087, 960, and 808 cm^–1 correspond to the antisymmetric stretching vibration of Si–O–Si, the bending vibration absorption of –OH and the symmetric stretching vibration of Si–O bond, respectively, further confirming the existence of SiO₂ around UCNPs. Besides the SiO₂ layer coating outside UCNPs, CeO₂–COOH nanoparticles distribute uniformly on the surface of UCNPs@SiO₂-NH₂ (Figure 1C,D) after amidation reaction, which has been confirmed by HRTEM and indicates the successful synthesis of CeO₂ modified UCNPs@SiO₂ (UCNPs@SiO₂@CeO₂) preliminarily. A relatively obvious characteristic peak at 1377 cm^–1 belonging to the internal vibration of amide bonds and a peak at 1726 cm^–1 belonging to stretching vibration of C = O in the carboxyl group can be found in the FTIR spectrum of amino-modified UCNPs@SiO₂ (UCNPs@SiO₂-NH₂) and CeO₂–COOH, respectively, and a new peak at 1637 cm^–1 in UCNPs@SiO₂@CeO₂ which belongs to the stretching vibration of C = O in amide bond confirms the successful amidation between UCNPs@SiO₂-NH₂ and CeO₂–COOH (Figure 1B). As a result, the cubic CeO₂ with fluorite structure (JCPDS 34-0394), tetragonal LiYbF₄: Tm³⁺ (JCPDS 23-0371), and LiYbF₄:Tm³⁺@LiYF₄ (UCNPs) (JCPDS 17-0874) can be observed in the XRD patterns of UCNPs@SiO₂@CeO₂ (Figure 1A), confirming that the UCNPs@SiO₂@CeO₂ are successfully synthesized. When the feed ratio between CeO₂–COOH and UCNPs@SiO₂-NH₂ increases, more CeO₂ nanoparticles attach to the surface of UCNPs@SiO₂ (Figure S3C–F). Furthermore, the analysis of XPS shows that both Ce³⁺ and Ce⁴⁺ exist in the CeO₂ cubes, which is very important for the antioxidation property of CeO₂ and endows CeO₂ with excellent wound healing promotion ability for chronic wounds (Figure 1E). PCL membranes loaded with UCNPs@SiO₂@CeO₂ can be obtained by an electrospinning method which exhibit uniform diameter and are not influenced by the addition of UCNPs@SiO₂@CeO₂ compared with pure PCL membranes (Figure S4). The durability of nanoparticles in PCL fibers was tested by soaking the membranes in PBS and incubating at 37°C in a shaker. As shown in Figure (S5), the fiber membranes have no significant change and obvious exfoliated nanoparticles in PBS under physiological conditions, indicating the dressing have good stability.

The optical properties of these nanoparticles are studied through UV–vis diffuse reflectance spectra (UV–vis DRS) and upconversion luminescence (UCL) spectrum tests (Figure 1F,G). UV–vis DRS results show that UCNPs almost have no absorption in the range of 350–700 nm, while the absorption wavelength of CeO₂ can range from the UV region to 450 nm. After modifying UCNPs with CeO₂, UCNPs@SiO₂@CeO₂ have absorption in the UV region as well which endows them with the possibility to be excited by UV light (Figure 1F). Figure (1G) shows the emission spectra of nanoparticles under the excitation of 980 nm NIR laser. Under 980 nm NIR laser irradiation, UCNPs show obvious emission at around 350 nm. But the emission intensity of both UCNPs@SiO₂ and UCNPs@SiO₂@CeO₂ (1-1), that is, the mass ratios between UCNPs@SiO₂-NH₂ and CeO₂–COOH are controlled to be 1:1, decreases obviously, especially for UCNPs@SiO₂@CeO₂ (1-1). The reason for UCNPs@SiO₂ is that the strong Si–O stretching outside UCNPs can affect the emission intensities through multi-phonon relaxation,^16,17 and the existence of CeO₂ around UCNPs can absorb UV light which further reduces the emission intensity of UCNPs@SiO₂@CeO₂ visibly.⁴⁴ It is well-known that ROS can influence cellular metabolism and cause damage to bacteria without inducing drug resistance. The ability of UCNPs, CeO₂, and UCNPs@SiO₂@CeO₂ (1-1) to produce ROS under 980 nm NIR irradiation in vitro was evaluated by 2′-7′- DCFH-DA assay and the intensity of peaks reflect the level of ROS. As shown in Figure (1H), the fluorescence spectra of the solutions containing UCNPs or CeO₂ have no obvious change before and after 980 nm NIR irradiation. However, for the UCNPs@SiO₂@CeO₂ (1-1), the intensity of the peak increases gradually with the increase of irradiation time, which proves that CeO₂ outside the UCNPs are triggered by the UV light (300–400 nm) that emitted by UCNPs under 980 nm NIR irradiation to produce ROS for PDT. A possible mechanism is that the UV light emitted by UCNPs under the irradiation of 980 nm NIR can be absorbed by CeO₂ on the surface of UCNPs to stimulate the electrons in CeO₂ from valence bands to conduction bands, and the left electron holes in the valence bands can interact with H₂O on the surface of UCNPs@SiO₂@CeO₂ to generate ROS for further applications (Figure 1I).

The antibacterial property of membranes

Based on our design that the UV light emitted by the UCNPs under 980 nm NIR can trigger CeO₂ to produce ROS to resist bacteria, the antibacterial properties of different naparticles against both S. aureus and E. coli were assessed by the plate coating method to acquire optimal intensity of irradiation. As shown in Figures (S6 and S7), when the irradiation intensity is 0.8 W cm^–2, none of the groups has antibacterial effects. As shown in Figures (S8 and S9), the UCNPs@SiO₂@CeO₂ composites exhibit better PDT antibacterial effect compared with other control groups when the power density is 1 W cm^–2 and the irradiation time is 5 min. When the irradiation intensity is 1.2 W cm^–2, the bacterial death due to the thermal effect of 980 nm NIR itself, which interferes with the antibacterial effect of PDT and even influences the cell viability in our subsequent experiments (Figures S10 and S11). Thus, 1 W cm^–2 is chosen as the final irradiation intensity. The fiber membranes are used as final dressings for wound healing, so the antibacterial performances of membranes against both S. aureus and E. coli at different irradiation times were further explored based on 1.0 W cm^–2. As shown in Figure (2A,B), all the experimental groups exhibit no obvious antibacterial ability without 980 nm NIR irradiation (the time of irradiation is 0 min), but when the irradiation time reaches 10 min, the groups treated with UCNPs@SiO₂@CeO₂/PCL membranes begin to perform antibacterial effects, especially groups of UCNPs@SiO₂@CeO₂ (2-1)/PCL (the mass ratios between UCNPs@SiO₂-NH₂ and CeO₂–COOH are controlled to be 2:1) and UCNPs@SiO₂@CeO₂ (1-1)/PCL. When the irradiation time is 15 min, the antibacterial effects become more obvious, but the UCNPs and CeO₂ loading PCL groups reveal the little antibacterial property. The reason is that, the ROS can be produced under the irradiation of 980 nm NIR to resist the bacteria. In this process, UV light emitted by the UCNPs under irradiation of 980 nm NIR can further trigger CeO₂ to release ROS. In addition, the inherent antibacterial capability of CeO₂ and UCNPs (emitted UV light) make them show little antibacterial ability, respectively. For the UCNPs@SiO₂@CeO₂/PCL membranes, the emitted UV light could stimulate CeO₂ modified on the surface of UCNPs to produce ROS, which is more lethal to bacteria. However, the survival number of colonies has no obvious correlation with the loading capacity of CeO₂. Combined with the TEM results in Figure (S3), UCNPs@SiO₂ can be completely coated by CeO₂ when the amount of loaded CeO₂ is enough, which will weaken the excitation effect of UV light on the peripheral CeO₂. Therefore, continuing increase of CeO₂ loading not only has no obvious improvement of antibacterial performance, but also causes certain biological toxicity.⁴⁵ As a kind of polymer with good biocompatibility, PCL has no response to 980 nm NIR light, and pure PCL membranes exhibit no antibacterial ability no matter with or without 980 nm NIR irradiation. Thus, compared with the antibacterial results of the nanoparticles, the use of PCL fiber membranes as dressings can absorb part of the heat, which is more beneficial to the application of wound dressing for diabetic wound healing.

Before and after 980 nm NIR treatment, the changes of bacteria morphology growing on UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes are characterized by the FESEM to further confirm the prominent antibacterial ability of UCNPs@SiO₂@CeO₂ (Figure 2C). Compared with the morphology of S. aureus and E. coli without 980 nm NIR treatment, the cell membranes of the two bacteria are severely damaged and no complete morphology can be found out after 980 nm NIR irradiation. These results indicate that UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes possess excellent photodynamic antibacterial capability under 980 nm NIR irradiation due to the production of ROS.

Cytotoxicity and morphology of the cell attached on a membrane

The biocompatibility of PCL membranes loaded with different concentrations of nanoparticles (0.5, 1, 2, 4, and 6 mg ml^–1) was tested by an MTT method using L929 cell lines (mouse fibroblast cell lines) (Figure 3A). Both UCNPs/PCL and CeO₂/PCL membranes exhibit good biocompatibility, even if the concentration of the nanoparticles is as high as 6 mg ml^–1. After combining UCNPs with CeO₂, the biocompatibility of UCNPs@SiO₂@CeO₂ with different mass ratios shows some differences. The L929 cells cultured with UCNPs@SiO₂@CeO₂ (2-1) or UCNPs@SiO₂@CeO₂ (1-1) still maintain a desired cell viability. The UCNPs@SiO₂@CeO₂ (1-2)/PCL (the mass ratios between UCNPs@SiO₂-NH₂ and CeO₂–COOH are controlled to be 1:2) has obvious cytotoxicity from the concentration of 4 mg ml^–1, and the UCNPs@SiO₂@CeO₂ (1-3)/PCL (the mass ratios between UCNPs@SiO₂-NH₂ and CeO₂–COOH are controlled to be 1:3) exhibit slight toxicity when the concentration at 0.5, 1, and 2 mg ml^–1, and it has a significant influence to cell viability when at the concentration of 4 and 6 mg ml^–1 (< 60%). Live–dead assay is further used to confirm the biocompatibility of the subsequently used PCL membranes loaded with different nanoparticles. Only few dead cells (red dots) can be detected for the L929 cells growing on the pure PCL membranes, which verify the well-known biocompatibility of PCL polymer (Figure S12). Similarly, few red dots can be found out for the L929 cells growing on the CeO₂/PCL, UCNPs/PCL, and UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes, further indicating the good biocompatibility of this system.

Cytoskeleton staining method is implemented to observe the morphology of L929 cells growing on the membranes. As shown in Figure (3B), L929 cells can spread out very well not only on pure PCL membranes, but also on the other membranes with different nanoparticles, including on the CeO₂/PCL, UCNPs/PCL, and UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes. This result demonstrates that there is no any influence on the growth status of cells after adding nanoparticles into PCL, which can be further confirmed by the FESEM images (Figure S13 and Figure 3C). The FESEM images show that L929 cells adhere to the membranes very well with spindle morphology, and the antennas of them contact with each other.

Test of the ability to remove reactive oxygen species

DCFH-DA is the most commonly fluorescent dye to detect the level of ROS in cells. After penetrating cell membranes, DCFH-DA is hydrolyzed by esterase to form 2′,7′-DCFH, which would be further oxidized by the ROS in cells into fluorescent 2′,7′-DCF. Due to the strong oxidizing property of H₂O₂, cells treated with H₂O₂ only (control group) exhibit high fluorescence intensity (Figure 3D). As UCNPs do not have antioxidation ability, adding UCNPs into the system cannot impair the oxidation effect of H₂O₂ to cells. Based on the well-known antioxidant ability of CeO₂, both CeO₂ and UCNPs@SiO₂@CeO₂ can prevent DCFH from being oxidized to DCF. As a result, the cells treated with CeO₂ nanoparticles or UCNPs@SiO₂@CeO₂ hardly reveal fluorescence at 525 nm, confirming the coating of CeO₂ on UCNPs endows UCNPs@SiO₂@CeO₂ with excellent antioxidant property, and the different weight ratios between UCNPs and CeO₂ do not influence the antioxidant ability obviously. ROS scavenging experiment in vitro preliminarily confirms the attractive antioxidant ability of UCNPs@SiO₂@CeO₂, which is very important to resist inflammatory response and promote healing of chronic wounds. Based on the results of antibacterial property, cell toxicity, and ROS removing ability, UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes would be chosen for the subsequent in vivo experiment.

Wound healing

Diabetic SD rats are used as animal models to detect the promotion ability of the as-prepared membranes to diabetic chronic wounds (Figure 4). In Figure (4A), on the first 3–5 days, both control and UCNPs/PCL groups exhibit the lowest healing rate no matter with or without 980 nm NIR irradiation. Because of the inherent antibacterial ability of CeO₂, CeO₂/PCL group reveals little promotion ability to chronic wounds whether with 980 nm NIR irradiation or not. As UCNPs@SiO₂@CeO₂ can produce ROS under 980 nm NIR irradiation to protect wounds from growing of bacteria, wounds treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes with 980 nm NIR irradiation exhibit the best healing rate at the first 3–5 days. Without 980 nm NIR irradiation, UCNPs@SiO₂@CeO₂ can only exhibit the inherent antibacterial ability of CeO₂, so that wounds treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes without 980 nm NIR irradiation represent similar wound healing rate as the CeO₂/PCL group. As time goes on, due to the special microenvironment, the excess ROS around wounds is caused by diabetes restrains the healing of chronic wounds. Based on the excellent antioxidant ability of CeO₂, both CeO₂/PCL and UCNPs@SiO₂@CeO₂ (1-1)/PCL groups can alleviate oxidative stress and help wounds transform from the inflammatory stage to the proliferation stage to promote wound healing no matter with or without 980 nm NIR irradiation. Combining both the PDT and antioxidant capability, wounds treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes with 980 nm NIR irradiation reveal the best healing rate, and the wounds almost repair completely after 15 days with little epidermis lost. The wound healing rate and remaining wound area are calculated by Image J, and the results are consistent with above results (Figure 4B,C).

To further investigate the effect of UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes (1-1/PCL) with PDT antibacterial property for the healing of infected wounds, the experiments with infection of S. aureus are carried out as Figure (4D). As shown in Figure (4E), the infected wounds are created successfully after 2 days of inoculating S. aureus, and the wound area in 1-1/PCL+NIR group begins to exhibit smaller than those in other groups on day 3. Over time, the wounds treated with both 1-1/PCL and irradiation of 980 nm NIR light keep a better healing trend than others obviously. The result of the residual number of bacteria in wound through different treatments is shown in Figure (4F). On day 5, the number of colonies significantly decreases in 1-1/PCL+NIR group compared to other groups. On day 7, the number of colonies in 1-1/PCL and 1-1/PCL +NIR groups is much less than that in control and control+NIR groups, indicating that the wound treated by 1-1/PCL+NIR group which possess the PDT antibacterial capacity can kill bacteria during the process of irradiation, and the inherent antibacterial ability of CeO₂ also plays an important role in the system. On day 10, the colonies almost disappeared because of the healing of wounds. The corresponding viability of S. aureus is shown in Figure (4G), which further certificates that the 1-1/PCL can efficiently inhibit the growth of bacteria under 980 nm NIR irradiation.

H&E staining and Masson's trichrome staining are carried out to further confirm the promotion ability of the as-prepared membranes (Figure 5). The boundary between the old and new born tissues around the wounds is marked by dark blue dotted lines (Figure 5A,B). A clear interface can be identified from day 8 for all the groups except for the control group without 980 nm NIR irradiation. It can be seen from the results that wounds treated by CeO₂/PCL and UCNPs@SiO₂@CeO₂ (1-1)/PCL no matter with or without 980 nm NIR irradiation reveal more new tissue and less inflammatory response on day 8, indicating the antibacterial ability in the first few days and continuous oxidation resistance can promote wound healing together and the new born tissues have formed on day 8. Most of all, the tissue of wounds treated by UCNPs@SiO₂@CeO₂ (1-1)/PCL group with 980 nm NIR irradiation shows the best state because of the damage of generated ROS to bacteria around wounds. Blood vessels can be found out on day 15 for the control and UCNPs/PCL groups both with and without 980 nm NIR irradiation. While for the wounds treated with CeO₂/PCL and UCNPs@SiO₂@CeO₂ (1-1)/PCL, both blood vessels and skin appendages, such as hair follicles, can be observed in the area of new tissues on day 15 no matter with or without 980 nm NIR irradiation, confirming the antioxidant ability of CeO₂ to promote chronic wound healing. Furthermore, mature epithelial structure even forms for the wounds treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL with 980 nm NIR irradiation after 15 days due to both the excellent antibacterial ability and the antioxidant property. The enlarged images on day 15 (Figure S14, circled in green in Figure 5) show that, in the groups of CeO₂/PCL and UCNPs@SiO₂@CeO₂ (1-1)/PCL, skin epidermis grows more completely and maturely. In addition, initial follicles can be observed from wounds treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL membranes with 980 nm NIR irradiation. Masson's trichrome staining is the most common used method to detect collagen deposition, and the density and order of collagen fibers can reflect the regeneration of new skin tissues during wound healing (Figure 5C,D). As shown in Figure (5C,D), the proportion on the right of the red dotted line is significantly higher than that on the left, which indicates the formation of granulation tissue during the remodeling process. At the same time, myofibroblasts gradually decrease, along with the gradual increase of collagen which finally deposits to form dense collagen fibers. On day 15, the collagen fibers are more densely and orderly packed of the wounds treated with UCNPs@SiO₂@CeO₂ (1-1)/PCL with 980 nm irradiation. The proportion of accumulated collagen fibers in other groups is smaller, and the density and order of collagen fibers are lower. Both H&E staining and Masson's trichrome staining show that UCNPs@SiO₂@CeO₂/PCL membranes exhibit the best effect of PDT and antioxidant capability to promote chronic wound healing with 980 nm NIR irradiation, and 980 nm NIR can be used as the switch to control photodynamic antibacterial ability, showing great potential for clinic.

CONCLUSIONS

A novel UCNPs@SiO₂@CeO₂/PCL membrane was successfully prepared through the amidation reaction and electrospinning method. The membranes possess excellent biocompatibility, PDT, and antioxidant activity. In this system, the combination of UCNPs and CeO₂ endows UCNPs@SiO₂@CeO₂/PCL membranes with attractive photodynamic antibacterial property to both S. aureus and E. coli under 980 nm NIR irradiation, which is very important to resist bacterial growth around chronic wounds at the early stage. Moreover, the reversible conversion between Ce³⁺ and Ce⁴⁺ in CeO₂ nanoparticles endows UCNPs@SiO₂@CeO₂/PCL membranes with excellent ability to alleviate oxidative stress around chronic wounds, which further promote the transformation of wounds from the inflammatory stage to the proliferative stage to accelerate chronic wound healing. Both in vitro and in vivo experiments confirm that modifying CeO₂ outside UCNPs endows UCNPs@SiO₂@CeO₂/PCL membranes with so attractive photodynamic antibacterial ability and outstanding antioxidant capability that can effectively promote chronic wound healing. Due to the good biosafety, controllability, and effectiveness in promoting chronic wound healing of UCNPs@SiO₂@CeO₂/PCL, it has great potential for the application in the treatment of chronic wounds.

ACKNOWLEDGMENTS

We gratefully acknowledge the financial support from the National Natural Science Foundation of China (21971117, 22004032, 52003124, 21771156), the Functional Research Funds for the Central Universities, Nankai University (63186005, ZB19500202), the State Key Laboratory of Rare Earth Resource Utilization (RERU2019001), the National Key R&D Program of China (No. 2017YFA0208000, 2021YFA1202400, 2021YFA1501101), the Outstanding Youth Project of Tianjin Natural Science Foundation (20JCJQJC00130), the Key Project of Tianjin Natural Science Foundation (20JCZDJC00650), China Postdoctoral Science Foundation (2020M680862), Natural Science Foundation of Tianjin City (21JCQNJC00280), the 111 Project (No. B18030) from China. We also thank the Haihe Laboratory of Sustainable Chemical Transformations for financial support.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

Xinyun Zhai and Yaping Du conceived and designed the study. Tengfei Ma carried out most of the experiments and analyzed the data. Mengdie Jin carried out a part of the experiments. Xinyun Zhai, Yongkang Huang, and Mengzhen Zhang performed a part of in vivo experiment. Tengfei Ma and Xinyun Zhai co-wrote the manuscript. Yaping Du, Xinyun Zhai, Tengfei Ma, Xiaoli Zhao, and Haobo Pan revised the manuscript.