Intraoperative application of an antioxidant nanoparticle-hydrogel targeting microglia regulates neuroinflammation in traumatic brain injury

Abstract

Microglia play a critical role in neuroinflammation, a key secondary injury mechanism following traumatic brain injury (TBI). The colony-stimulating factor 1 receptor (CSF-1R) inhibitor PLX5622 has shown promise in suppressing neuroinflammation by depleting microglia, but it lacks specificity in targeting microglia at the injury site. To overcome this limitation, we developed PLX5622 nanoparticles functionalized with the CAQK peptide for lesion-specific targeting and combined them with a hydrogel (GelMA-PPS) that possesses potent reactive oxygen species (ROS) scavenging capabilities. This nanoparticle-hydrogel drug delivery system (GelMA-PPS/P) significantly enhanced the delivery efficiency and therapeutic efficacy of PLX5622 in TBI treatment. Localized administration of this system effectively depleted microglia at the injury site, suppressed neuroinflammation, and reduced the release of inflammatory cytokines. Its ROS scavenging ability was also validated in vitro and in vivo. Together, these effects synergistically improved neurological function recovery in TBI mouse models. This innovative strategy offers a comprehensive and targeted approach to managing neuroinflammation after TBI, providing a promising avenue for advancing TBI therapies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-025-03682-7.

Introduction

Traumatic brain injury (TBI) represents a critical global public health concern with growing prominence in contemporary society. The substantial prevalence of TBI, coupled with its considerable economic burden, underscores its significance. Severe TBI cases, in particular, are associated with alarmingly high rates of mortality and disability [1]. The severity of TBI is due not only to its high incidence and mortality rate but also to its long-term impact on patients [2]. Addressing these challenges has become an urgent priority for the global medical community, necessitating advancements in clinical treatment strategies to reduce mortality, alleviate disability, and enhance the quality of life for patients [3, 4].

TBI encompasses both primary and secondary brain injuries. Secondary brain injury refers to the secondary damage that occurs following the initial traumatic event, driven by a cascade of pathological processes. These processes contribute to ongoing neurological impairment in patients suffering from cranial trauma [5]. Currently, clinical treatments for TBI are relatively limited. In addition to conservative approaches, the focus for severe cases remains primarily on early-stage surgical interventions, with a lack of targeted treatments for secondary damage. Therefore, addressing secondary brain injury is both a key focus and a significant challenge in the clinical management of TBI [6, 7]. One of the core mechanisms of secondary brain injury is neuroinflammation, which involves the activation and release of various inflammatory cells and mediators. The main characteristics of neuroinflammation include the excessive release of inflammatory cytokines, disruption of the blood-brain barrier (BBB), and exacerbation of cerebral edema [8]. Additionally, neuroinflammation is closely associated with reactive oxygen species (ROS) and the oxidative stress they cause. The excessive accumulation of ROS leads to lipid peroxidation, protein damage, and DNA damage, which further exacerbates cellular injury and inflammatory responses [9]. This interaction between inflammation and oxidative stress creates a vicious cycle of secondary brain injury, which results in in more severe brain tissue damage and functional impairment [10].

Neuroinflammation after TBI involves powerful and complex interactions between central and peripheral cells, as well as soluble factors. TBI can lead to early activation of resident microglia in the central nervous system, accompanied by peripheral neutrophil recruitment and subsequent infiltration of lymphocytes and monocyte-derived macrophages [11]. Moreover, the successive expression or secretion of proinflammatory and anti-inflammatory mediators can promote or terminate the neuroinflammatory response after TBI, respectively. Among these mediators, chemokine-related signaling pathways can lead to the activation and recruitment of immune cells toward the site of injury [12]. The neuroinflammatory response after TBI is a double-edged sword: it has beneficial effects, such as promoting debris clearance and regeneration by mediating neuronal death and progressive neurodegeneration, and harmful effects, such as excessive secretion of cytokines and chemokines leading to BBB disruption, prolonged inflammatory processes, and further induction of chronic neurodegeneration. Identifying strategies to “seek benefits and avoid harm” in the neuroinflammatory response remains a challenge when treating TBI patients and is a focus of current research and exploration [13].

To reasonably control the neuroinflammatory response after TBI, the current option is to use microglial depletion therapy, which inhibits colony stimulating factor 1 receptor (CSF-1R) [14]. Under physiological conditions, microglia are the only cell type in the central nervous system that expresses CSF-1R, which is an important pathway through which individuals maintain the survival and proliferation of microglia [15]. In models brain injury, such as TBI and stroke, PLX5622 can clear microglia and inhibit neuroinflammation, ultimately improving the prognosis of brain injury model mice [16]. However, PLX5622 still has several limitations in practical application that hinder its clinical use: (1) PLX5622 lacks the ability to selectively deplete microglia at the site of injury; (2) PLX5622 not only depletes microglia in the central nervous system but also leads to the depletion of peripheral bone marrow-derived monocyte-macrophages; (3) prolonged oral administration of PLX5622 can significantly affect the gut microbiota; (4) the first-pass effect of oral administration reduces the bioavailability of PLX5622; and (5) patients with severe TBI require surgical intervention, and early postoperative fasting poses a significant limitation for the oral administration of PLX5622 [17–19].

To further explore the role of PLX5622 in the consumption of microglia and the inhibition of neuroinflammation, as well as to minimize the above-described limitations of PLX5622, we introduced liposome nanoparticles and light-curing hydrogel technology to encapsulate PLX5622. Moreover, we used structural optimization and material interaction to target these nanoparticles to the site of TBI and exert antioxidant effects [20]. We first prepared PLX5622 nanoparticles and modified their surface with cysteine–alanine–glutamine–lysine (CAQK) peptides, which can specifically target the injury site. These nanoparticles were then encapsulated within an antioxidant hydrogel (GelMA-PPS) to form a nanoparticle-hydrogel drug delivery system (GelMA-PPS/P). GelMA (gelatin methacryloyl) is a biodegradable hydrogel precursor derived from gelatin through methacryloyl (MA) modification. Due to its excellent biocompatibility and tunable mechanical properties, GelMA has been widely used in biomedical fields such as tissue engineering and drug delivery. Therefore, we selected GelMA as the hydrogel matrix in this study. After traumatic brain injury surgery, GelMA-PPS/P was injected into the injury site, where it transformed from a liquid state to a gel state under blue light irradiation. The polyphenylene sulfide (PPS) in GelMA-PPS/P responds to and scavenges ROS, reducing ROS levels while simultaneously promoting the breakdown of the hydrogel and releasing PLX5622 nanoparticles. The CAQK peptides on the surface of the PLX5622 nanoparticles bind to chondroitin sulfate proteoglycans (CSPGs) that are specifically expressed after injury, enabling the nanoparticles to accumulate at the trauma site. This allows for depletion of microglia in the injury area and the inhibition of neuroinflammation (Fig. 1). We believe that this intraoperatively applied GelMA-PPS/P nanoparticle-hydrogel drug delivery system can reduce oxidative stress following TBI while exerting neuroprotective effects by specifically depleting microglia in the injury area and modulating neuroinflammation.

Fig. 1.

Schematic diagram of the intraoperative Application of Antioxidant Nanoparticle-Hydrogel Targetting Microglia Regulates the Neuroinflammation in TBI. (A) The assembly process of GelMA-PPS/P. (B) GelMA-PPS/P is injected into the injury site during surgery and self-assembles into a gel that adheres to the injured area. (C) GelMA-PPS/P responds to the post-TBI microenvironment, degrades, and releases PLX5622 nanoparticles to deplete microglia. (D) GelMA-PPS/P reduces neuroinflammation by both scavenging ROS and depleting microglia through the release of the CSF-1R inhibitor PLX5622. This figure was created by Figdraw

Materials and methods

Synthesis and characterization of CAQK-DSPE-PEG₂₀₀₀

DSPE-PEG₂₀₀₀-MAL was dissolved in N, N-dimethylformamide, and CAQK and triethylamine were added to dissolve completely. The reaction was carried out at room temperature for 12 h. The reaction solution was dialyzed in pure water for 24 h (cut-off molecular weight, 1000 Da) and freeze-dried to obtain the product. A ¹H nuclear magnetic resonance spectroscopy (¹H-NMR) was performed for validation. During purification, the reaction mixture was transferred into a dialysis bag with a molecular weight cutoff of 1000 Da and dialyzed against ultrapure water for 24 h to remove unreacted small-molecule CAQK peptide. After dialysis, the solution inside the dialysis bag was collected and lyophilized to obtain the purified CAQK-DSPE-PEG₂₀₀₀ conjugate.

Preparation and characterization of PLX562 nanoparticles

Lecithin (0.1 mg), CAQK-DSPE-PEG₂₀₀₀ (0.04 mg), DSPE-PEG₂₀₀₀ (0.1 mg), PLX5622 (0.1 mg), and PLGA (0.2 mg) were completely dissolved in dimethyl sulfoxide (DMSO, 200µL) at 25 ℃. The PLGA used in this study was purchased from MedChemExpress (USA), with a molecular weight of 10,000 Da. The mixture was dripped at a constant speed into PBS (800 µL) which was continuously stirred, and stirred at 25 ℃ for 2 h to synthesize PLX5622-CAQK nanoparticles. The solution was purified and dialyzed before storage or further application.

A dynamic light scattering (DLS) detector (Zetasizer Pro, Malvern Panalytical, UK) was used for measuring the size distributions and zeta potential of PLX5622-CAQK nanoparticles at 25 °C. The average size automatically calculated by the DLS software is the intensity-weighted harmonic mean diameter, which is a widely used standard parameter for nanoparticle size characterization. To evaluate structural stability, the nanoparticles were stored in PBS at room temperature from Day 0 to Day 14 and analyzed at designated time points.

PLX5622-CAQK nanoparticle solutions on Day 0 and Day 14 were placed onto copper grids and air-dried. Then it was stained by uranium acetate and its micromorphology was observed by transmission electron microscopy (TEM) (Hitachi, Japan).

The encapsulation efficiency and content were detected by high-performance liquid chromatography (HPLC, LD-20AD, Shimadzu, Japan). The formula of drug loading and encapsulation efficiency are as follows:

Drug loading = (weight of drug loaded)/(weight of nanoparticles) × 100% = 16.2%.

Encapsulation efficiency = (weight of drug loaded)/(weight of drug added) × 100% = 87.6%.

To determine the drug content in PLX5622-CAQK nanoparticles, the purified nanoparticles were diluted with the mobile phase (0.1% trifluoroacetic acid in water: acetonitrile = 55:45, v/v) and vortexed for 2 min to disrupt the nanoparticle structure. The mixture was then sonicated at room temperature (25 °C) using an ultrasonic cleaner (40 kHz, 100 W) for 10 min to ensure complete drug release, followed by HPLC analysis. The chromatographic conditions were as follows: an Ultimate Plus-C18 column (4.6 mm × 150 mm, 5 μm); mobile phase: 0.1% trifluoroacetic acid in water and acetonitrile (55:45, v/v); flow rate: 1.0 mL/min; UV detection at 254 nm; injection volume: 5 µL; column temperature: 30 °C.

Synthesis and characterization of PPS₁₂₀-MA

1,8-Diazabicyclo [5.4.0] undec-7-ene (DBU) in dry tetrahydrofuran (THF) was degassed for 30 min in a dried and nitrogen- flushed 50 mL round bottle flask. The reaction mixture temperature was lowered to 0 °C and degassed for 30 min. In this flask, a degassed solution of 1-butanethiol in THF was added dropwise and reacted for 30 min. Then, freshly distilled and degassed propylene sulfide monomer was added to the reaction mixture, and the temperature was maintained at 0 °C for 2 h. The reaction mixture was quenched by the addition of 2-iodoethanol and stirred overnight. The next day, the polymer solution was filtered and further purified by three precipitations into cold methanol before vacuum-drying. Finally, the colorless viscous polymer PPS₁₂₀-OH was obtained. PPS₁₂₀-OH was dissolved in triethylamine and dichloromethane. Then, freshly distilled MA chloride was added dropwise at 0 °C. The mixture was stirred overnight under argon. The resulting products were dialyzed (cut-off molecular weight, 2000 Da) and lyophilized to give PPS₁₂₀-MA. ¹H-NMR was performed for validation. Infrared (IR) spectra and ultraviolet-visible (UV-Vis) absorption spectroscopy were employed to further characterize PPS₁₂₀-MA. IR analysis was performed using a spectrometer equipped with an attenuated total reflectance (ATR) accessory (UV-2600, Shimadzu, Japan). Prior to measurement, the dried sample was placed directly onto the ATR crystal and gently pressed to ensure optimal contact. Spectra were collected in the range of 4000–500 cm⁻¹ with a resolution of 4 cm⁻¹ and 32 scans per sample. Background correction was conducted before sample measurement. The resulting spectra were used to identify functional groups and verify the copolymer structure. UV-Vis absorption spectra were acquired using a UV-Vis spectrophotometer (SP-2, PerkinElmer, USA). PPS₁₂₀-MA was first dissolved in chloroform at an appropriate concentration, with the aid of ultrasonication or magnetic stirring to ensure complete dissolution. The solution was then filtered through a 0.45 μm membrane to remove any undissolved particulates. Measurements were carried out using a quartz cuvette with a path length of 10 mm, and chloroform was used as the reference solvent. The absorption spectra were recorded over a wavelength range of 200–600 nm to analyze the electronic structure and conjugation characteristics of the sample.

Preparation of GelMA-PPS/P

GelMA (50 mg), PPS₁₂₀-MA (20 mg, dissolved in 40 µL DMSO in advance) and lithium phenyl-2,4,6-trimethylbenzoylphosphinate (5 mg) were added in nanoparticle solution (1 mL) and swirled thoroughly in dark condition. In this study, the GelMA-PPS/P hydrogel was crosslinked under 405 nm light to form a solid gel. To minimize potential phototoxicity and ensure biosafety, the irradiation parameters were rigorously controlled, with the power density maintained below 50 mW/cm² and the exposure time limited to a maximum of 30 s. The GelMA was purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (catalog number M752743, China).

Scanning electron microscopy (SEM) of GelMA-PPS/P

The cured hydrogel was frozen in liquid nitrogen for 5 min. Then it was freeze-dried for 48 h in a freeze-dryer. An appropriate amount of sample was placed on a plate, and the truncated surface was used for gold spraying treatment. The surface was observed by SEM (Hitachi, Japan).

In vitro drug release

GelMA-PPS/P was collected and placed in a 15 mL tube every 5 mL volume. According to the literature, we used artificial cerebrospinal fluid (CSF) (3525/25 ML, Tocris) to serve as a simulated CSF [21]. Then, 10 mL of PBS, H₂O₂ (500 µM), artificial CSF, and CSF from post-operative TBI patients were added separately. The tubes were shaken at 37 °C at 70 rpm. A 100 µL portion of the hydrogel immersion solution was taken at different time points (6 h, 12 h, 24 h, 2 days, 3 days, 4 days, 5 days, 6 days, 7 days, 8 days, 10 days, 12 days, and 14 days) and stored. 100 µL of the corresponding solution was replenished. Our drug release curve actually quantifies the PLX5622 contained within the nanoparticles released from the hydrogel, thereby indirectly reflecting the release behavior of the nanoparticles. The quantification method for PLX5622 was the same as described above.

Characterization of hydrogel

The rheological properties of hydrogel were determined at 37 ℃ by using a dynamic Discovery HR-2 rheometer (TA Instruments, USA). Hydrogel samples (400 µL) were added to a 40 mm parallel plate using a syringe, and the gap of the plate was set to 500 µm. Data of elastic modulus (G’) and viscosity modulus (G”) were recorded with dynamic strain frequency sweep (0–100 rad/s, strain 5%) and time sweep (5 rad/s, strain 1%, 120 s). A strain sweep test was performed using a rheometer at a constant frequency of 1 Hz, with the strain ranging from 0.01 to 1000%, to record the variation of G’ and G” as a function of strain. This test was conducted to determine the linear viscoelastic region (LVR), structural stability, and yield behavior of the hydrogel. All measurements were carried out using a parallel-plate geometry. Compression testing of freeze-dried hydrogel samples was conducted using a universal testing machine (Kezhun Test Instruments, China). The samples were prepared in a cylindrical shape and compressed axially at a constant speed (e.g., 1 mm/min) at room temperature. Stress–strain curves were recorded, and the slope within the linear strain region (typically 0–10% strain) was used to calculate the compression modulus, which reflects the hydrogel’s resistance to deformation. For the swelling test, pre-weighed dried hydrogel samples (W₀) were immersed in PBS at 37 °C. At designated time points (e.g., 1 h, 2 h, 4 h, 12 h, 24 h, up to 120 h), the samples were removed, surface moisture was blotted off, and the wet weight (Wₜ) was recorded. The swelling ratio was calculated as follows: Swelling rate = (Wₜ - W₀)/W₀ × 100%. All experiments were performed in triplicate, and the average values were used to plot the swelling kinetics curve. Thermal stability of the dried hydrogel samples was assessed using thermogravimetric analysis with a thermal analyzer (Netzsch, Germany). The samples were heated from room temperature to 800 °C at a rate of 10 °C/min under nitrogen atmosphere. The remaining weight% at different temperatures was recorded to evaluate the material’s thermal decomposition profile and stability.

1,1-diphenyl-2-picryl-hydrazyl radical (DPPH•) and 2,2’-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) radical cation (ABTS⁺•) scavenging rate

After 500 µL of hydrogel precursor solution was light-crosslinked into a gel within a test tube, 500 µL of 20 µmol/L DPPH solution (MedChemExpress, USA) was added. The mixture was vortexed and incubated in the dark on a shaker for 30 min. The supernatant was then collected, and its absorbance at 517 nm was recorded as Ai. As a control, 500 µL of light-crosslinked hydrogel was mixed with 500 µL of 50% ethanol, incubated under the same conditions, and the absorbance of the supernatant at 517 nm was recorded as Aj. A mixture of 500 µL DPPH solution and 500 µL of 50% ethanol (without hydrogel) was used as the blank control, and the absorbance was recorded as A₀.The DPPH• scavenging rate (%) was calculated using the following formula: DPPH• scavenging rate (%) = [1 - (Ai - Aj)/A₀] × 100%.

After 500 µL of hydrogel precursor solution was light-crosslinked into a gel within a test tube, 500 µL of ABTS solution (Beyotime, China) (prepared by dissolving 3.84 mg ABTS and 0.66 mg K₂S₂O₈ in 0.5 mL ultrapure water in a 1.5 mL microcentrifuge tube, incubated in the dark for 12–16 h, and then diluted with absolute ethanol to 50 mL) was added. The mixture was vortexed and incubated on a shaker at room temperature in the dark for 30 min. The supernatant was collected, and its absorbance at 734 nm was recorded as Ai. As a control, 500 µL of light-crosslinked hydrogel was mixed with 500 µL of distilled water and incubated under the same conditions; the absorbance of the supernatant at 734 nm was recorded as Aj. A mixture of 500 µL ABTS⁺• solution and 500 µL distilled water (without hydrogel) was used as the blank control, and the absorbance was recorded as A₀. The ABTS⁺• scavenging rate (%) was calculated using the following formula: ABTS⁺• scavenging rate (%) = [1 - (Ai - Aj)/A₀] × 100%.

Cell culture

A microglial cell line (BV2), a hippocampal neuronal cell line (HT22), and an astrocyte type I clone cell line (C8) were cultured for in vitro experiments. The culture medium was Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 1% penicillin/streptomycin solution and 10% fetal bovine serum (FBS). The three types of cells were cultured in a humidified cell incubator with 5% CO₂ at 37 ℃.

In vitro uptake experiment

The PLX5622-CAQK nanoparticles were synthesized as described above, with 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate (DiI) (MedChemExpress LLC, China) added during synthesis to enable fluorescent tracing. Meanwhile, nanoparticles without the addition of CAQK component were prepared as control. BV2 cells were seeded in 12-well plates at 1 × 10⁶ per well on the day before treatment. Cells of the activated group were incubated with H₂O₂ (800 µM) for 2 h at 37 ℃, while the normal group cells were in original DMEM. After wash with PBS for 3 times, the cells were cocultured separately with DMEM containing PLX5622-CAQK nanoparticles and PLX5622 nanoparticles for 4 h 37 ℃. After nanoparticle incubation, the cells were fixed at room temperature with 4% paraformaldehyde for 15–20 min, followed by washing 3 times with PBS. The fixed cells were then incubated in 300 nM DAPI staining solution under light-protected conditions for 5–10 min, followed by another three PBS washes. The image was observed under a fluorescence microscope (DiI, excitation 550 nm, emission 565 nm; DAPI, excitation 350 nm, emission 470 nm). The mean fluorescence intensity was calculated using ImageJ software.

In vitro 2, 7-dichlorodihydrofluorescein diacetate (DCFH-DA) assay

In the in vitro experiments, since it was not possible to fully replicate the in situ light-crosslinking environment of the surgical setting, the hydrogels were pre-crosslinked before being used for co-culture with cells. BV2, HT22 and C8 cells were separately seeded in 12-well plates at 1 × 10⁶ per well on the day before treatment. Cells were cocultured with light-crosslinked (solidified) hydrogels or PBS as a control. Then cocultured with H₂O₂ (800 µM) for 2 h at 37 °C. After wash with PBS for 3 times, cells were incubated with 0.1% DCFH-DA, Beyotime, China) in serum-free medium at 37 ℃ for 20 min. Cells were subsequently washed for 3 times with PBS to remove free DCFH-DA. Then the image was observed under a fluorescence microscope at 488 nm excitation and 525 nm emission, and the mean fluorescence intensity was calculated using ImageJ software.

In vitro cytotoxicity assay

BV2, HT22 and C8 cells were separately seeded in 96-well plates at 1 × 10⁵ per well on the day before treatment. Cells were cocultured with light-crosslinked (solidified) hydrogels. The Control group were in DMEM without any hydrogel. After wash with PBS for 3 times, cells were added 10 µL/well Cell Counting Kit-8 (CCK-8) reagent (Vazyme, China) for 1 h. The absorption was measured at 450 nm using a microplate reader.

Controlled cortical impact (CCI) animal models

C57BL6/J Mice (male, 6–8 weeks old) were purchased from the GemPharmatech LLC, China. We chose to establish a CCI model on mice, and the detailed method is as follow. All animal experiments complied with the ARRIVE guidelines and were carried out in accordance with the National Research Council’s Guide for the Care and Use of Laboratory Animals [22]. Mice were anesthetized with isoflurane and maintained during the process. The skull was fully exposed under aseptic operation, and a 5 mm diameter craniotomy was performed at a point midway between the lambda and bregma sutures and laterally midway between the central suture and left temporalis muscle. After removing the bone flap, PinPiont™ PCI3000 Precision Cortical Impactor (Hatteras Instruments, USA) was used to construct a serve CCI model, and the parameters are as follow: impactor piston diameter, 3 mm; impact velocity, 5.5 m/s; impact depth, 2.5 mm; duration, 100 ms. After CCI, the contusion tissue was removed and hemostasis was performed. Finally, the scalp was sutured and mice were allowed to recover on a heating mat to maintain a normal body temperature. The mice were randomly divided into 5 groups. The Sham group mice only received craniotomy and scalp suture without CCI. The TBI group mice only received CCI without using any hydrogel. The GelMA-PPS, GelMA/P, and GelMA-PPS/P group mice were injected with 50 µL of corresponding hydrogel at the lesion and solidified before scalp suture.

RNA-Sequencing

On the 7th day after TBI, mice were anesthetized and the cortex tissue around the TBI site was collected. Further RNA sequencing was conducted by Shanghai BMbios Co. Ltd (China), with the following steps. After extraction from the cortex tissue, the total RNA was purified and evaluated for quality. The RNA library was prepared and raw data was obtained on the sequencing platform. The FastQC tool was used to evaluate the data quality. The analysis of differences in expression genes and enrichment analysis were conducted. The corresponding data are provided in the Supplementary material 2.

Hematoxylin and Eosin (HE) staining of major organs

HE staining of major organs was performed on the 21 st day after TBI. Mice were anesthetized, and after rapid left ventricular infusion of saline and 4% paraformaldehyde, the heart, liver, spleen, lungs and kidneys were removed. These organs were prepared into paraffin sections and stained with corresponding staining solutions. Finally, the sections were observed with a visible light microscope.

Dihydroethidium (DHE) staining

On the 7th day after TBI, mice of each group were anesthetized with isoflurane, and after rapid left ventricular infusion of saline, the entire brain tissue was removed on ice. The brain tissue was rapidly cooled and frozen, and sliced into 20 μm frozen sections. DHE dye (MedChemExpress LLC, China) was used for staining to display ROS levels. The image was observed under a fluorescence microscope at 535 nm excitation and 610 nm emission.

Positron emission tomography-computed tomography (PET-CT)

On the 3rd, 7th, 14th, and 21 st day after TBI, ¹⁸F-DPA-714 PETCT imaging was performed to continuously monitor neuroinflammation in vivo. The radiosynthesis of ¹⁸F-DPA-714 was performed according to previous literatures, and the precursors were purchased from Huayi Technology. Co. Ltd, China. The ¹⁸F was produced by an accelerator (Sumitomo, Japan) and a FN multifunctional module (GE healthcare, USA) was used for synthesis. After intravenously injection of ¹⁸F-DPA-714 (100 µCi), the mice were placed in a warm condition for 45 min. The mice were anesthetized and the static PET imaging lasted for 10 min using a small animal PET-CT (Siemens, Germany). According to previous studies [23], the Standardized Uptake Value (SUV) is commonly used to evaluate the metabolic activity or binding capacity of tissues for radiotracers.

Enzyme linked immunosorbent assay (ELISA)

After anesthesia, the mouse brain tissue was rapidly isolated and the cortex around the TBI site was collected. The cortex tissue was homogenized thoroughly in the lysis solution, and the supernatant was collected after centrifugation. The protein concentration was quantitatively analyzed using a BCA protein detection kit (Vazyme, China). The contents of inflammatory factors and ROS markers at different time points after TBI were analyzed using an ELISA kit (Jonlnbio, China). The contents of IL-1β, IL-6, and TNF-α were detected on the 7th day after TBI and the contents of superoxide dismutase (SOD), glutathione peroxidase (GSH-px) and malondialdehyde (MDA) were detected on the 7th day after TBI.

Brain tissue sections Preparation and staining

At different time points after TBI, mice of each group were anesthetized with isoflurane, and after rapid left ventricular infusion of saline and 4% paraformaldehyde, the entire brain tissue was removed. Then brain tissue embedding in paraffin were prepared into 20 μm sections with a paraffin slicing machine. These paraffin sections underwent immunostaining, with the following steps. The sections were dewaxed, and placed in citric acid repair solution for antigen repair. The sections were incubated in 3% H₂O₂ at room temperature for 25 min, followed by incubation in BSA solution at room temperature for 30 min. The sections were incubated with different primary antigen solution of ionized calcium-binding adapter molecule 1(IBA-1, 1:2000, ab178847, Abcam, UK), NeuN (1:1000, ab104224, Abcam, UK), glial fibrillary acidic protein (GFAP, 1:1000, 3670 s, CST, USA) at 4 ℃ overnight. After washing with PBS for 3 times, the slices were incubated at room temperature for 1 h in the corresponding secondary antibody solution. Next, a DAB solution was used for color development, and hematoxylin solution was used for staining cell nuclei. Nissl staining and HE staining were directly performed using the corresponding working solutions after dewaxing. Finally, the sections were dehydrated and sealed, and images were observed with a visible light microscope.

Magnetic resonance imaging (MRI)

We performed MRI examination on the 7th day after TBI to monitor changes in brain edema. The detailed parameters of the 7.0T micro-magnetic resonance scanner (Bruker, Bio Spec 70/20) machine are as follows: repetition time = 2500 ms, echo time = 36 ms, layer thickness = 1 mm, field of vision = 20 × 20 mm, and image matrix = 256 × 256. Subsequently, we used ImageJ to calculate the brain edema volume.

Evans blue (EB) experiment

On the 7th day after TBI, we intravenously injected EB (45 mg/kg, Sigma, USA) into mice of each group following a 2 h circulation. After anesthesia and left ventricular saline infusion, the injured cerebral hemisphere was separated and weighed. The injured hemisphere was placed in 1 mL of 30% trichloroacetic acid solution and homogenized. After incubation at 70 ℃ for 72 h, the mixture was centrifuged for 10 min at 10,000 g. 200 µL supernatant was collected, and its absorption was measured at 620 nm using a microplate reader.

Measurement of brain water content

After anesthesia, the brain tissue of mice was separated and the injured cerebral hemisphere was isolated to measure the wet weight. Then the brain tissue was dried at 80 °C for 3 days, and the dry weight was measured. The formula of brain water content is as follow: brain water content = (wet weight - dry weight)/wet weight × 100%.

Neurological function assessment

On the 0th, 3rd, 7th, and 14th day after TBI, the modified Neurological Severity Scores (mNSS) ranging from 0 (normal) to 18 (severest) were performed to evaluate the abilities of locomotor, sensory, balance, and reflexes.

To assess motor coordination after injury, mice underwent Rota-rod training for 3 consecutive days prior to surgery, following a published protocol [24]. On day 1, the maximum speed was set at 20 rpm (accelerating from 4 to 20 rpm over 150 s, total duration 5 min); on day 2, 30 rpm (4 to 30 rpm over 150 s, 5 min); and on day 3, 40 rpm (4 to 40 rpm over 150 s, 5 min). Animals unable to maintain performance for 5 min after training were excluded. After training, mice underwent CCI surgery to establish the TBI model. On the 3rd, 7th, and 14th day after CCI, the Rota-rod test was used to evaluate the motor ability. Mice of each group were placed on a uniformly rotating rod to maintain their posture until they fell off, and the latency to fall was recorded with a maximum value of 5 min. A longer latency to fall suggests better motor function.

The Barnes maze was performed from the 21 st to 28th day after TBI to evaluate the long-term learning and memory abilities of mice. Placed in the center of a circular platform, mice instinctively avoided bright light and started to search for the dark target hole among the 20 equidistant holes at the edge of the platform. The monitoring of animal movement trajectories and the acquisition and analysis of behavioral parameters were performed using a camera positioned above the maze and connected to a video tracking system (TopScan systems, CleverSys, Inc., CSI). The training phase lasted for 6 days from the 21 st day, and memory test was conducted after an interval of one day.

Statistical analysis

All of the data are presented as the means ± standard deviation (SD). All statistical analyses were performed using SPSS version 21 (IBM, USA) and plots were created using GraphPad Prism version 8 (GraphPad Software, USA). Statistical significance was analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test. In neurological function assessment, to compare differences among groups over multiple time points, a two-way ANOVA was performed, with “group” and “time” as fixed factors and individual animals as repeated measures. Post-hoc multiple comparisons were conducted using the Tukey test. A P-value < 0.05 was considered statistically significant, with ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001 indicating different levels of significance.

Results

Synthesis and characterization of GelMA-PPS/P

We crosslinked the CAQK peptide with DSPE-PEG₂₀₀₀-MAL to synthesize CAQK-DSPE-PEG₂₀₀₀ and verified its structure via ¹H-NMR. As shown in Fig. S1A, the synthesis process and chemical structures of CAQK, DSPE-PEG₂₀₀₀-MAL, and the final conjugate CAQK-DSPE-PEG₂₀₀₀ are illustrated. Regarding the specific chemical linkage between CAQK and DSPE-PEG₂₀₀₀-MAL, we performed ¹H-NMR analysis comparing the CAQK-DSPE-PEG₂₀₀₀ conjugate with its raw materials (CAQK and DSPE-PEG₂₀₀₀-MAL). The results showed that the characteristic proton signal of the alkene double bond in the maleimide moiety of DSPE-PEG₂₀₀₀-MAL at 6.98 ppm (H-e) disappeared in the conjugate, indicating that this structure served as the Michael acceptor during the reaction. In addition, the characteristic proton signal of the thiol group in CAQK at 2.59 ppm also disappeared, further confirming that the thiol group of CAQK reacted with the maleimide group of DSPE-PEG₂₀₀₀-MAL through a Michael addition reaction, forming a stable thioether bond in the final conjugate. The amide N-H proton signals of the CAQK-DSPE-PEG₂₀₀₀ product were assigned as follows: 8.65 ppm (m, 0.67H, H-a), 8.28 ppm (s, 0.72H, H-f), 8.14 ppm (m, 0.93H, H-b), 7.73 ppm (s, 1H, H-c), 7.20 ppm (s, 0.72H, H-d’), 6.77 ppm (s, 0.67 H, H-d’). Based on peak integration, the area of the amide N-H signal at 8.28 ppm (H-f) corresponding to DSPE-PEG2000 was 0.67, while the total area of the CAQK amide N-H signals at 7.20 ppm and 6.77 ppm (H-d) was 0.67 × 2 = 1.34, resulting in an approximate ratio of 1:2, consistent with a complete reaction at a 1:1 molar ratio of DSPE-PEG₂₀₀₀-MAL and CAQK, confirming the reaction’s completeness and expected stoichiometry (Fig. S1B). The PLX5622 nanoparticles were self-assembled via the organic solvent injection method (Fig. 2A). In this study, PLX5622, a hydrophobic small-molecule drug, was encapsulated within the hydrophobic PLGA core of the nanoparticles, while the outer layer was composed of lecithin, DSPE-PEG₂₀₀₀-CAQK, and DSPE-PEG₂₀₀₀ to enhance nanoparticle stability and targeting capability. The lipid–polymer hybrid nanoparticle synthesis strategy employed in this study was adapted and refined from previously reported methods [25–27], and is particularly suitable for the efficient encapsulation of hydrophobic drugs.

Fig. 2.

Synthesis and characterization of GelMA-PPS/P. (A) Schematic diagram of the construction of PLX5622 nanoparticles. (B) Particle size (a) and Zeta potential of nanoparticles (b). (C) TEM images of PLX5622 nanoparticles at Day 0 and Day 14 (scale bar = 100 nm). (D) SEM images of the structure of different gel components (scale bar = 200 μm). (E) Gelation process of GelMA-PPS and GelMA-PPS/P in vitro. (F) Gelation process of GelMA-PPS/P in vivo. (G) In vivo targeting of GelMA-PPS/P in a mouse TBI model (scale bar = 200 μm)

The average diameter of the nanoparticles were 114.62 ± 1.58 nm, with a low polydisperdsity index (PDI = 0.1423). The zeta potential was − 0.9729 mV (Fig. 2B). The PDI below the conventional monodispersity threshold of 0.2 indicated a relatively narrow particle size distribution and good uniformity. The TEM images indicated that the PLX5622 nanoparticles exhibited a spherical structure with good monodispersity. The size of the nanoparticles shown in the TEM images was consistent with the above results. We stored the nanoparticles in PBS at room temperature and retested them in vitro. Both the TEM results and the scattering detector results indicate that the nanoparticles had good structural stability from Day 0 to Day 14 (Fig. 2C and S2). In terms of drug-loading performance, the HLPC results revealed that the drug loading of PLX5622 nanoparticles loaded with PLX5622 was 16.2%, and the encapsulation efficiency was 87.6%. These results demonstrate that we successfully constructed the PLX5622 nanoparticles needed for the experiment, and they exhibited excellent stability.

Based on previous literature reports [21, 22], we synthesized a PPS polymer with a degree of polymerization of 120 and further methacrylated it to obtain PPS₁₂₀-MA. As shown in Fig. S3A, the structure was characterized by ¹H NMR (CDCl₃, 400 MHz). The peaks at δ = 6.10 and 5.56 ppm (a, b) correspond to the allylic protons of the MA groups, indicating the successful grafting of MA moieties onto the ends of the PPS backbone. The signals at δ = 2.65–3.20 ppm (e, f) are assigned to methylene or methine protons adjacent to sulfur atoms, δ = 1.40 ppm (d) corresponds to the aliphatic methylene protons in the PPS backbone, and δ = 2.01 ppm (c) is attributed to methylene protons adjacent to the MA groups. In the IR spectra, the absorption peaks at 3000–2800 cm⁻¹ are associated with alkyl C-H stretching vibrations, indicating the presence of organic groups in the polymer. The peaks at 1600–1500 cm⁻¹ correspond to the C = C stretching vibrations of aromatic rings, confirming the aromatic structure of PPS. The peaks at 1200–1000 cm⁻¹ are attributed to C-S stretching vibrations, characteristic of PPS, while the peak at 1700 cm⁻¹ reflects the C = O stretching vibration, indicative of the ester group in MA. These features collectively confirm the successful copolymerization of PPS and MA (Fig. S3B). In the UV-Vis absorption spectra, a strong absorption peak is observed in the 200–300 nm range, and the polymer exhibits significant ultraviolet absorption, typically related to π-π* transitions of aromatic rings and n-π* transitions of conjugated structures. The absorption characteristics in this region further support the presence of aromatic rings and conjugated systems, indicating that the polymer contains conjugated structures capable of interacting with ultraviolet light (Fig. S3C). These results confirm the successful synthesis of the target structure PPS₁₂₀-MA and provide reactive sites for subsequent UV-induced free-radical copolymerization with GelMA. Through methacrylate double-bond crosslinking, the PPS segments can be effectively integrated into the GelMA network, endowing the composite hydrogel with enhanced antioxidant properties and tunability.

With the help of a photoinitiator, we successfully solidified a mixture of GelMA, PSS₁₂₀-MA and PLX5622 nanoparticles in a hydrogel under 405 nm light. The SEM results of the different types of nanoparticle-hydrogel systems are shown in Fig. 2D. Although the hydrogel pores are relatively large, SEM images reveal a typical three-dimensional interconnected network structure that can provide spatial confinement and a localized retention environment for the nanoparticles. This, together with the physical crosslinking of polymer chains within the gel matrix, contributes to a sustained drug release profile. Moreover, our study employed an “in situ encapsulation” strategy, in which the nanoparticles were co-delivered with the precursor solution directly into the lesion site and crosslinked under light irradiation to form the hydrogel in situ. This approach effectively enhances local drug retention and enables controlled release at the injury site. The light curing gelation of the GelMA-PPS/P is directly shown in Fig. 2E. In the left reagent bottle, the unexposed hydrogel was liquid, and its good fluidity was conducive to its ability to cover the surface of the lesion and ensure localized effects. In the middle and right reagent bottles, the naive hydrogels and those loaded with PLX5622 nanoparticles were solidified after being irradiated with 405 nm light. The light-curable gelling properties of the hydrogels remained stable before and after loading of the nanoparticles. Furthermore, we established a TBI model and injected GelMA-PPS/P into the injury site. Under blue light, the hydrogel formed in situ, adhering to the damaged area (Fig. 2F).

We labeled the PLX5622 nanoparticles with Dil and monitored their release at the injury site. As shown in Fig. 2G, the region with intense red fluorescence did not show DAPI staining, corresponding to the location of the GelMA-PPS/P locally injected into the injury area. In addition, no obvious Dil red fluorescence accumulation was observed in the contralateral hemisphere, suggesting that GelMA-PPS/P was primarily confined to the injection site without diffusing to the opposite side. Moreover, red fluorescence was also observed around DAPI-positive nuclei in the injured hemisphere, further indicating that the nanoparticles within GelMA-PPS/P accumulated in the injured brain tissue.

As shown in Fig. S4, we placed GelMA-PPS/P in different liquid environments and measured the release of PLX5622 to evaluate the drug release efficiency from the hydrogel over time. In this study, we acknowledge that the artificial CSF used differs from biological CSF in both composition and complexity. The application of artificial CSF has certain limitations, as it cannot fully replicate the intricate biochemical environment of native CSF in vivo. However, obtaining CSF from healthy human donors is associated with stringent ethical approvals, technical challenges, and limited donor availability. Even animal-derived CSF is difficult to collect in sufficient quantities and is subject to significant inter-individual variability, making it unsuitable for reproducible in vitro studies. Therefore, in our in vitro drug release experiments, we adopted an artificial CSF formulation based on previously study [21]. This formulation offers well-defined composition, controlled sourcing, and high reproducibility, allowing us to evaluate the baseline release characteristics of the drug under standardized conditions and to perform reliable comparisons across experimental groups. As shown in the Fig. S4, it was observed that the release profiles of GelMA-PPS/P in PBS and artificial CSF exhibited comparable trends. In the first two days, slow release was observed, and no obvious release was observed at the subsequent time points. The cumulative release was maintained below 10%. GelMA-PPS/P exhibited the highest release rate within the first 3 h in both H₂O₂-stimulated conditions and TBI CSF. In the H₂O₂ group, the release remained elevated from 3 to 12 h before plateauing. In the TBI CSF group, the presence of various matrix metalloproteinases gradually degraded the GelMA matrix, resulting in a more sustained release profile, although the release rate still peaked during the initial 0–3 h window. This release kinetics profile, jointly driven by elevated ROS and pathological factors, aligns well with the temporal requirements for early neuroinflammatory intervention post-TBI. We have incorporated the corresponding release curve data into the revised manuscript and further elaborated on the time-responsiveness and pathophysiological relevance of the GelMA-PPS/P system in the treatment of TBI.

Rheological analysis was used to verify the mechanical properties of GelMA-PPS/P and evaluate its flow and deformation under external conditions under dynamic strain frequency scanning and time scanning, respectively. The results are shown in Fig. S5A and B. When G’ exceeds G’’, the system can be considered to approximate a solid colloid. Over time, the G’ values of different hydrogels exceeded their corresponding G” values. Furthermore, the mechanical strength of the gels was relatively constant, which indicates that they maintained a stable gel state. In dynamic strain frequency scanning, the G’ values of different hydrogels also exceeded their corresponding G” values in the overall frequency range, which still indicates that the solid structure of each group remained stable. As shown in Fig. S5C, all four groups maintained a stable G’ in the low-strain region, indicating good network integrity. Notably, the GelMA-PPS and GelMA-PPS/P groups exhibited higher G’ values and the widest LVR, suggesting superior crosslinking density and structural stability. The stress-strain curves demonstrated typical nonlinear elastic behavior across all groups (Fig. S5D). Among them, the GelMA-PPS/P group displayed the highest maximum stress and compressive modulus, indicating a denser network structure and enhanced mechanical strength (Fig. S5E). All hydrogels exhibited rapid water uptake and swelling within the first 24 h, followed by a gradual approach to saturation. The GelMA and GelMA/P groups showed higher swelling ratios, reflecting a relatively loose network structure, whereas the GelMA-PPS and GelMA-PPS/P groups exhibited lower swelling rates and equilibrium swelling ratios, indicative of higher crosslinking density (Fig. S5F). As shown in Fig. S4G, all samples displayed major weight loss between approximately 200–350 °C, with residual mass remaining at around 20–30% above 350 °C.

Assessment of the antioxidant properties of GelMA-PPS/P using DPPH• and ABTS⁺• assays

To evaluate the free radical scavenging ability of the different hydrogel formulations, DPPH• and ABTS⁺• radical scavenging assays were performed. As shown in Fig. S6, both GelMA-PPS and GelMA-PPS/P groups exhibited superior scavenging efficiency compared to the GelMA and GelMA/P groups in both assays, indicating that the incorporation of PPS endowed the hydrogels with ROS responsiveness and antioxidant activity. These results confirmed the free radical scavenging capability of the GelMA-PPS/P hydrogel and provided supportive evidence for its subsequent application in cellular and in vivo experiments.

PLX5622 nanoparticles can be taken up by microglia following injury

We used BV2 cells to evaluate the ability of PLX5622 nanoparticles to be taken up by microglia in vitro. The fluorescence results after 4 h of incubation are shown in Fig. 3A. The blue fluorescence of DAPI represents the nuclei of BV2 cells, whereas the red fluorescence of DiI represents PLX5622 nanoparticles. Under normal conditions, BV2 cells were not treated with hydrogen peroxide and did not produce large amounts of CSPGs. Due to the lack of targets for CAQK, no significant aggregation of red fluorescent particles was observed in either the Control group or the CAQK group. In the H₂O₂-activated treatment group, CSPGs provided specific binding sites for nanoparticle aggregation, and an increased uptake of red-labeled nanoparticles was observed around cells in the CAQK group. We also performed quantitative analysis of DiI fluorescence intensity across groups, and the group treated with H₂O₂ and co-incubated with CAQK-modified nanoparticles exhibited the highest fluorescence intensity (Fig. S7).

Fig. 3.

PLX5622 nanoparticles can be specifically taken up by microglia after injury. (A) In vitro uptake experiment of PLX5622 nanoparticles. The cells used in this experiment were BV2 (scale bar = 50 μm). GelMA-PPS/P scavenged ROS in vitro. (B) Fluorescence microscopy images of BV2, HT22, and C8 cells, and green fluorescence indicates the intracellular ROS levels (scale bar = 50 μm). (C) Quantitative analysis of fluorescence intensity measured by a microplate reader (n = 3). GelMA-PPS/P selectively exerts cytotoxic effects only on microglia in vitro. (D) The cytotoxic effects of different drug treatments on BV2, HT22, and C8 cells. Data are presented as the means ± SD. ^*p < 0.05 and ^***p < 0.001

GelMA-PPS/P scavenged ROS in vitro

The thioether groups in PPS can be oxidized by ROS while simultaneously depleting ROS [28]. As a result, GelMA-PPS/P can reduce ROS levels and inhibit oxidative stress. We adopted a DCFH-DA fluorescence method to evaluate the antioxidant effects of different formulations of GelMA-PPS/P on major types of brain cells in vitro. We used hydrogen peroxide stimulation to induce ROS production in BV2 cells, HT22 cells and C8 cells. We then coincubated these cells with GelMA-PPS/P extract to evaluate its anti-ROS effects. The DCFH-DA fluorescence images of the BV2, HT22, and C8 cell lines clearly differ. In the PBS group, a high level of ROS was generated because these cells only received PBS containing hydrogen peroxide. The GelMA/P group also expressed a high level of ROS because these cells were treated with a hydrogel that did not contain PPS. In the GelMA-PPS and GelMA-PPS/P groups, the antioxidant effect of PPS significantly inhibited the production of ROS, which showed that the GelMA-PPS/P hydrogel alleviated oxidative damage in a variety of in vitro neural cell lines (Fig. 3B and C).

GelMA-PPS/P selectively exerts cytotoxic effects only on microglia in vitro

Our goal was to selectively target and eliminate microglia at the injury site and avoid any cytotoxic effects on other neural cells, such as neurons and astrocytes. We used the CCK-8 cytotoxicity assay to further evaluate the effect of GelMA-PPS/P on cell proliferation across different in vitro cell lines. As shown in Fig. 3D, the viability of BV2 cells in the GelMA-PPS group (hydrogel without PLX5622 nanoparticles) was comparable to that in the Control group, indicating that the hydrogel itself had no effect on microglial viability. In contrast, the GelMA/P and GelMA-PPS/P hydrogels containing PLX5622 nanoparticles significantly reduced BV2 cell viability. This finding suggests that this approach is expected to further control neuroinflammatory responses at the in vivo level. The viability of the HT22 and C8 cells in each group was similar to that in the Control group, which shows that GelMA-PPS/P does not kill neurons or astrocytes in vitro.

GelMA-PPS/P has good biocompatibility

Biocompatibility is a crucial factor in determining the potential clinical application of GelMA-PPS/P. As shown in Fig. S8, histopathological analysis of major organs (heart, liver, spleen, lungs, and kidneys) was performed in TBI mice treated with different interventions to evaluate biocompatibility. No significant pathological changes were observed in any of the treatment groups.

GelMA-PPS/P downregulates the transcriptome related to ROS damage and neuroinflammation

We collected brain tissues from the mice in the Sham, TBI, and GelMA-PPS/P groups on the 7th day after injury for RNA-sequencing analysis. The results of the analysis of the differences in the expression of genes are shown in Fig. 4A-C. Compared with the Sham group, the TBI group had 3919 genes whose expression was significantly upregulated after injury, including genes important for inflammation activation and ROS damage mechanisms, such as Aif1, Casp1, Il6ra, Nox4, Nlrp3, and Tnf. Compared to the TBI group, 1294 genes were significantly downregulated following GelMA-PPS/P administration. These genes were related to the pathways involved in inflammation activation and ROS damage mentioned above. We further compared the expression differences between the GelMA-PPS/P group and the Sham group. We focused on genes that are important for the mechanisms of inflammatory activation and ROS damage mentioned above. We found that most of these genes showed no differences in expression, which indicates that GelMA-PPS/P intervention can inhibit inflammatory activation and ROS damage.

Fig. 4.

GelMA-PPS/P down-regulates transcriptome related to ROS damage and neuroinflammation. (A) Volcano plot of differentially expressed genes between the TBI group and the Sham group. (B) Volcano plot of differentially expressed genes between the GelMA-PPS/P group and the TBI group. (C) Volcano plot of differentially expressed genes between the GelMA-PPS/P group and the Sham group. (D) Heatmap showing differentially expressed genes among the Sham, TBI, and GelMA-PPS/P groups. (E) GO and KEGG enrichment analysis between the TBI group and the Sham group. (F) GO and KEGG enrichment analysis between the GelMA-PPS/P group and the TBI group. (G) GO and KEGG enrichment analysis between the GelMA-PPS/P group and the Sham group

As shown in Fig. 4D, we selected several representative genes for heatmap analysis. The downregulation of neuron marker genes, such as Rbfox3 and Map2, after TBI was inhibited by the application of GelMA-PPS/P. The trend of genes related to inflammation activation and ROS damage was similar to that of the above gene volcano map. Further GO and KEGG enrichment analyses were performed (Fig. 4E-G). Compared with the Sham group, the TBI group showed significant enrichment of inflammation-related pathways such as NF-κB signaling, suggesting that TBI induces a robust neuroinflammatory response. When comparing the GelMA-PPS/P group with the TBI group, biological processes associated with oxidative damage, such as the NADPH oxidase complex, as well as inflammatory pathways like NF-κB, were also significantly enriched, indicating that GelMA-PPS/P may modulate these pathological processes. Notably, no such enrichment was observed when comparing the GelMA-PPS/P group to the Sham group, suggesting that the treatment may effectively reverse TBI-induced activation of oxidative stress and inflammatory pathways. These findings collectively indicate that GelMA-PPS/P likely exerts its therapeutic effects by regulating neuroinflammation and oxidative stress pathways following TBI.

GelMA-PPS/P can reduce microglia in the injured area and alleviate the inflammatory response

At different time points, we assessed the levels of activated microglia and neuroinflammation in the injury site of TBI mice after treatment via PET-CT, ELISA, and histological analysis (Fig. 5A). To further investigate whether GelMA-PPS/P can selectively eliminate microglia in the injured area after TBI, we innovatively used PET-CT to monitor microglial levels in the injury site of TBI mice following different treatments. The ¹⁸F-DPA-714 PET probe, as a ligand for translocator protein (TSPO), can specifically bind to TSPO sites on activated microglia, with TSPO being regarded as a marker of neuroinflammation [29]. Combining the radioactive tracer characteristics of 18 F with visualized and quantifiable in vivo imaging, ¹⁸F-DPA-714 PET-CT imaging is an ideal method for monitoring activated microglia and neuroinflammation in vivo. We selected this noninvasive, in vivo monitoring technique to dynamically observe changes in neuroinflammation at the injury site over time following the application of GelMA-PPS/P in a TBI model. The images of the Sham group revealed that under physiological conditions, SUV of the entire brain, including the cortex and cerebellum, were relatively low, indicating that neuroinflammation was low. After a TBI, the cortex SUV around the TBI site gradually increased, generally reaching a peak approximately 7 days after injury, after which it slowly decreased but remained at a high level on the 21 st day. These findings suggest that the neuroinflammatory response, as a secondary injury, gradually worsens and then decreases, which may lead to chronic inflammation. In the GelMA-PPS, GelMA/P, and GelMA-PPS/P groups, the increasing trend in the SUV value was alleviated, and the SUV in these groups significantly differed from that in the TBI group, especially on the 7th day. Specifically, the GelMA-PPS/P group had the lowest SUV value at all time points (Fig. 5B and C).

Fig. 5.

GelMA-PPS/P can effectively deplete microglia in the injured area and alleviate the inflammatory response. (A) Schematic diagram of PET-CT, ELISA, and histological analysis. (B) ¹⁸F-DPA-714 PET-CT images at different time points. (C) Quantitative analysis of ¹⁸F-DPA-714 PET-CT uptake (n = 3). (D) IBA-1 immunohistochemistry images on days 7 and 21 post-TBI in mice (scale bar = 50 μm). (E) Statistical analysis of IBA-1 positive cell counts (n = 3). (F) IL-1β, (G) IL-6 and (H) TNF-α levels of each group (n = 3). Data are presented as the means ± SD. ^*p < 0.05, ^**p < 0.01 and ^***p < 0.001

IBA-1 is closely linked to microglia and neuroinflammation [30]. IBA-1 is a cytoplasmic protein that is specifically expressed in microglia and plays a crucial role in the immune response of the nervous system [31]. When brain injury or neuroinflammation occurs, microglia become activated, leading to a significant increase in IBA-1 expression. Measuring IBA-1 levels through immunohistochemistry is a key method for quantifying microglial activity and assessing the extent of neuroinflammation in TBI. In the TBI group, the expression of IBA-1 rapidly increased after TBI, reached a high level approximately 7 days after injury, and slightly decreased on the 21 st day after injury, indicating possible chronic inflammation. This trend is consistent with previous literature reports and the PET-CT results mentioned above. In the GelMA-PPS, GelMA/P, and GelMA-PPS/P groups, the expression of IBA-1 was significantly reduced at these two time points, which suggests that the consumption of PLX5622 by microglia and the anti-ROS effect of PPS can reduce the neuroinflammatory response after TBI. Specifically, the expression of IBA-1 in the GelMA-PPS/P group was the lowest, indicating that the combination of these two components can play a more effective role (Fig. 5D and E).

Activated microglia exacerbate inflammation by releasing proinflammatory cytokines [32, 33]. On the 7th day after TBI, the levels of IL-1β, IL-6, and TNF-α in the cortex around the TBI site, as detected by ELISA, were significantly increased. Compared with those in the TBI group, the expression of inflammatory factors in the groups treated with different hydrogels was lower, and the expression levels in the GelMA-PPS/P group were the lowest, which suggested that GelMA-PPS/P has a specific anti-inflammatory effect and reduces the release of inflammatory factors (Fig. 5F-H).

We used the PET probe ¹⁸F-DPA714 and small animal PET-CT to monitor the inflammatory changes during the treatment of the TBI mouse model at multiple time points in vivo to investigate the protective effect of GelMA-PPS/P on neuroinflammation in the TBI mouse model. Moreover, combined with immunohistochemical analysis of the small glial cell marker IBA-1 and protein ELISA detection of major proinflammatory factors, the inhibitory effect of GelMA PPS/P on neuroinflammation was fully and comprehensively evaluated. After the application of the GelMA-PPS/P in TBI animal models, the nuclide uptake of ¹⁸F-DPA-714 PET/CT was reduced, the proliferation of IBA-1⁺ cells were inhibited, and the expression and secretion of various proinflammatory factors were reduced, all of which proved that it effectively consumed microglia and inhibited neuroinflammation.

GelMA-PPS/P exhibits antioxidant capabilities in vivo

DHE is a commonly used ex vivo to detect ROS. DHE can undergo oxidative reactions with ROS free radicals to form ethidium bromide, which further produces red fluorescence [34]. As shown in Fig. 6A and S9, after TBI, the red fluorescence signal increased at the TBI site. After treatment with the GelMA-PPS and GelMA-PPS/P hydrogels containing PPS, the signal intensity significantly decreased, as confirmed by quantitative analysis. These findings suggest that PPS can effectively inhibit the accumulation of ROS in vivo.

Fig. 6.

GelMA-PPS/P exhibits antioxidant capabilities in vivo. (A) DHE staining of brain tissue sections from each group (scale bar = 50 μm). (B) SOD, (C) GSH-px and (D) MDA of each group (n = 3). GelMA-PPS/P reduces BBB damage and cerebral edema. (E) Representative EB leakage images of each group. (F) Representative MRI images of each group. (G) Quantification of EB leakage of each group (n = 3). (H) Quantification of volume of hyperintensity of each group (n = 3). (I) Quantification of water content in brain tissue of each group (n = 3). Data are presented as the means ± SD. *p < 0.05, **p < 0.01 and ***p < 0.001

In addition to PPS directly scavenging ROS, we further assessed the endogenous antioxidant capacity of TBI mice following treatment. We selected several oxidative stress markers, namely, SOD, GSH-px, and MDA, to further evaluate their antioxidant effects in vivo. As shown in Fig. 6B-D, after TBI, the SOD and GSH-px levels in mouse brain tissue decreased, whereas the MDA levels increased, which is consistent with the secondary changes associated with ROS damage. After being treated with GelMA-PPS and GelMA-PPS/P hydrogels containing PPS, the SOD and GSH-px levels increased significantly, whereas the MDA levels decreased significantly, and this change was most pronounced in the GelMA-PPS/P group. These results indicate that GelMA-PPS/P exerts neuroprotective effects by both directly reducing oxidative stress through ROS scavenging and enhancing the endogenous antioxidant capacity.

GelMA-PPS/P reduces BBB damage and brain edema

Disruption of the BBB is closely linked to the development of brain edema after TBI [34]. The BBB is essential for maintaining brain homeostasis, but when damaged, its increased permeability allows proteins, fluids, and immune cells to leak into brain tissue, disrupting the brain microenvironment [35]. This disruption leads to fluid accumulation and worsened brain edema, which not only exacerbates the injury but can also impair neurological function and become life-threatening [36]. Neuroinflammation and ROS play key roles in exacerbating BBB damage after TBI. Activated microglia release proinflammatory cytokines that directly attack BBB endothelial cells, impairing their function [37]. ROS cause damage by oxidizing lipids, proteins, and DNA while also damaging tight junction proteins essential for BBB integrity. This damage increases BBB permeability, worsening brain edema and neural injury. Neuroinflammation and ROS form a vicious cycle, with inflammation increasing ROS production, which in turn worsens inflammation and BBB damage [38]. Once breached, more proinflammatory substances and ROS enter the brain, further exacerbating neuroinflammation and oxidative stress, ultimately leading to neuronal damage and worsening brain edema. Thus, reducing neuroinflammation and ROS is a critical strategy for protecting the BBB, reducing brain edema, and preventing further injury.

We used the EB assay to assess BBB damage in mice that received different treatments. Except for those in the Sham group, mice in all the other groups exhibited EB leakage on the 7th day after injury due to brain tissue and BBB damage. Specifically, the amount of EB was the lowest in the GelMA-PPS/P group, which showed that the GelMA-PPS/P hydrogels can effectively maintain the integrity of the BBB (Fig. 6E and G). On the 7th day after injury, the T2-weighted MR images of the mice revealed significantly high signal areas in the TBI group, indicating brain edema. In the GelMA-PPS, GelMA/P, and GelMA-PPS/P groups, the volume of the high-signal area was reduced. Specifically, the GelMA PPS/P group had the lowest value, indicating the ability of this hydrogel to prevent brain edema (Fig. 6H). Similarly, on the 7th day after injury, the TBI group presented the highest brain water content, indicating the most severe degree of brain edema. TBI mice treated with GelMA-PPS/P presented the lowest brain tissue water content (Fig. 6I).

The experimental results indicate that GelMA-PPS/P reduces excessive oxidative stress and neuroinflammation following TBI, attenuates BBB disruption and brain edema, which may contribute to mitigating secondary brain injury.

GelMA-PPS/P has neuroprotective effects in vivo

The NeuN gene is a biomarker of mature neurons and is commonly used for intuitive assessments of neuronal death or loss [39, 40]. Nissl bodies are basophilic granules that exist in the cytoplasm of neurons and can specifically display neurons and nerve fibers [41]. Therefore, we comprehensively used NeuN immunohistochemical staining and Nissl staining to evaluate neuronal damage. On the 7th day after injury, the number of NeuN⁺ cells in the TBI group sharply decreased, indicating severe neuronal death, which can often lead to a poor prognosis (Fig. 7A and B). In the GelMA-PPS, GelMA/P, and GelMA-PPS/P groups, the number of NeuN⁺ cells in each group increased compared with that in the TBI group. Specifically, this number was highest in the GelMA-PPS/P group and was closer to the number of mature neurons in the physiological state. In the Sham group, Nissl-stained cells were arranged neatly, with a normal morphology and uniform cytoplasmic staining. On the 7th day after injury, the number of Nissl-stained cells in the TBI group sharply decreased, and under high magnification, Nissl bodies in the cells were observed to shrink, decrease, or even disappear. Compared with those in the TBI group, the number of Nissl-stained cells in the GelMA-PPS/P hydrogel-treated group was greater, and the morphology was more complete. These experimental results suggest that the combination of PLX5622 and PPS can more effectively protect neurons (Fig. 7A and C).

Fig. 7.

GelMA-PPS/P has neuroprotective effects in vivo. (A) Representative NeuN, Nissl and GFAP images of each group (scale bar = 50 μm). Quantification of NeuN⁺ cell (B), Nissl cell (C) and GFAP⁺ cell (D) of each group (n = 3). (E and F) Representative HE images of each group on Day 7 and 21 (scale bar = 50 μm). (G and H) Quantification of lesion volume of each group on Day 7 and 21 (n = 3). Data are presented as the means ± SD. *p < 0.05, **p < 0.01 and ***p < 0.001

GFAP is a biomarker of reactive astrocytes. Therefore, we used GFAP immunohistochemical staining to characterize changes in astrocytes. On the 7th day after injury, the number of GFAP⁺ cells in the TBI group increased. In the GelMA-PPS, GelMA/P, and GelMA-PPS/P groups, the number of GFAP⁺ cells decreased compared with that in the TBI group, and the inhibitory effect was more pronounced in the GelMA-PPS and GelMA-PPS/P groups, indicating that GelMA-PPS/P can inhibit excessive proliferation of astrocytes (Fig. 7A and D).

We used HE staining to evaluate the loss of lesions and structural damage. ImageJ software was used to measure the areas of the injured and contralateral hemispheres separately, and the proportions of necrotic and lost contusions were calculated for quantitative comparison. In the Sham group, both hemispheres of the brain were symmetrical, and under high magnification, various types of cells in the cortex had intact structures and full cytoplasm, with tightly arranged nerve fibers. On the 7th day after injury, the impact injury affected the entire cortex and even the hippocampus, as shown in the images. HE staining of the TBI group revealed extensive loss of cortex and hippocampus tissue, accompanied by significant cell reduction and morphological abnormalities. As the damaged tissue further liquefied and necrotized, the degree of tissue loss on the 21 st day further increased compared with that on the 7th day, and this trend was reflected in all the experimental groups. Compared with that in the TBI group, the loss of injured tissue in each group treated with the hydrogel was alleviated at two time points. Among these groups, the loss of the GelMA-PPS/P group at two time points was the lowest, and the damage to the cortex and hippocampus was mild. The cells were arranged neatly, and the morphological damage was mild. These findings suggest that the combined use of PLX5622 and PPS can more effectively alleviate the necrosis of injured tissue and has good neural repair potential (Fig. 7E-H).

GelMA-PPS/P promotes neurological function recovery after TBI

Improving neurological deficits after TBI is of significant clinical importance [42]. We systematically evaluated motor function, as well as learning and memory abilities, in mice that received different treatments (Fig. 8A). The mNSS includes aspects such as movement, sensation, response, and balance, which can provide a relatively simple and comprehensive evaluation of residual symptoms after brain injury [43]. In the Sham group, the scores at each time point were good. In the TBI group, various behavioral abilities, primarily movement and sensation, were severely impaired. In the groups treated with different types of hydrogels, the mNSS at different time points tended to decrease, and the scores of the GelMA-PPS/P group were the lowest, which showed that the hydrogels effectively improved neurological function after TBI (Fig. 8B).

Fig. 8.

GelMA-PPS/P promotes neurological function recovery after TBI. (A) Schematic diagram of mNSS, rotarod test and Barnes maze (n = 5). (B) Quantification of mNSS at different time points. Details of the statistical analysis are provided in Table. S1. (C) Quantification of Rotarod at different time points. Details of the statistical analysis are provided in Table. S2. (D) Quantification of latency to target in learning phase. Details of the statistical analysis are provided in Table. S3. (E) Quantification of latency to target in memory test. (F) Quantification of distance moved in learning phase. Details of the statistical analysis are provided in Table. S4. (G) Quantification of atency to target in memory test. (H) The movement tracks of learning and memory test. Data are presented as the means ± SD. ^*p < 0.05, ^**p < 0.01 and ^***p < 0.001

The rotarod test can reflect an animal’s motor ability after injury, and a longer latency to fall suggests better motor function [44]. As shown in Fig. 8C, the mice in the Sham group maintained a good duration of nearly 5 min. After TBI, the ability to move is impaired, and the latency to fall is significantly reduced. In the groups that received different types of hydrogels, the latency to fall at different time points was improved, and that in the GelMA-PPS/P group was the highest, which showed that the hydrogel may improve the motor function of TBI mice.

The Barnes maze was used to evaluate the long-term learning and memory abilities of the mice after injury [45]. On the 21 st day after TBI, the mice in all the groups were trained and allowed to learn for 6 days. After one day had passed, a memory test was conducted on the 28th day. We chose the latency and distance traveled by the mice to reach the target hole for quantitative analysis. As shown in Fig. 8D-G, after 3 days of training, most of the mice in the Sham group consistently and rapidly reached the target hole. The latency to reach the target and distance moved were significantly impaired after TBI, and the mice were unable to reach the target hole effectively in the memory test. The typical movement tracks of learning and memory are shown in Fig. 8H. Among the groups subjected to different treatments, the GelMA-PPS/P group presented the most pronounced improvement, reflecting the recovery of spatial learning and memory ability, and the latency to target and distance moved in the memory test were closest to those in the normal physiological state.

Discussion

TBI constitutes a significant global public health challenge, characterized by high incidence, substantial economic burden, and severe clinical outcomes. Severe TBI is particularly associated with elevated mortality and long-term neurological disability [1]. These challenges highlight the urgent need for improved therapeutic strategies to mitigate mortality, reduce disability, and enhance long-term functional recovery [2]. TBI involves both primary mechanical damage and secondary injury driven by complex pathophysiological cascades [5]. Among these, neuroinflammation plays a central role, characterized by microglial activation, peripheral immune cell infiltration, excessive cytokine release, BBB disruption, and cerebral edema [46]. This inflammatory response is closely intertwined with oxidative stress, as the accumulation of ROS exacerbates cellular damage and amplifies inflammation, forming a vicious cycle that worsens neurological outcomes [20]. While neuroinflammation contributes to debris clearance and repair, its prolonged activation promotes chronic neurodegeneration. Current clinical interventions for TBI remain limited, with a lack of therapies specifically targeting secondary injury [6]. Thus, selectively modulating the neuroinflammatory response to mitigate its harmful effects while preserving its protective functions has emerged as a critical focus in the search for effective TBI treatments.

To modulate neuroinflammation after TBI, microglial depletion via CSF-1R inhibition has emerged as a promising strategy [47]. PLX5622, a CSF-1R inhibitor, effectively depletes microglia and attenuates neuroinflammation in preclinical models of TBI and stroke [48–50]. PLX5622 reduces microglia by inhibiting the CSF-1R signaling pathway, a mechanism that does not distinguish between pro-inflammatory (M1) and anti-inflammatory (M2) phenotypes, resulting in broad-spectrum depletion. While this approach may reduce both detrimental M1 and reparative M2 microglia, studies have shown that in the early phase after TBI, particularly in severe cases, M1 microglia rapidly predominate and drive neuroinflammatory cascades through the release of pro-inflammatory cytokines. Therefore, reducing overall microglial activation at this stage may help mitigate pathological damage. However, the clinical translation of PLX5622 remains limited due to several drawbacks, including lack of injury-site specificity, off-target depletion of peripheral monocyte-macrophages, disruption of gut microbiota with prolonged oral use, reduced bioavailability due to first-pass metabolism, and impracticality of oral administration in early postoperative TBI patients.

However, the clinical translation of PLX5622 remains limited due to several drawbacks, including lack of injury-site specificity, off-target depletion of peripheral monocyte-macrophages, disruption of gut microbiota with prolonged oral use, reduced bioavailability from first-pass metabolism, and impracticality of oral administration in early postoperative TBI patients.

To overcome the limitations of systemic PLX5622 administration and enhance its therapeutic efficacy, we developed a targeted nanoparticle-hydrogel delivery system (GelMA-PPS/P) for intraoperative application in TBI. PLX5622 was encapsulated in liposomal nanoparticles modified with CAQK peptides, which specifically bind to CSPGs upregulated at the injury site. CAQK is a small tetrapeptide with high tissue penetration and low immunogenicity, allowing it to distribute widely in vivo without eliciting significant immune responses. The primary target of CAQK is CSPGs, which become specifically exposed on the cell surface following injury, enabling CAQK to selectively target injured neural cells. CAQK was originally identified through in vivo phage display screening and has been shown to exhibit significant targeting capability for areas of neural injury [51–53]. These properties underlie the mechanism by which CAQK-modified nanoparticles in this study selectively accumulate at the injury site. CSPGs in the injured brain tissue are not secreted solely by microglia but result from the combined contribution of multiple cell types, leading to an overall upregulation of CSPGs in the injury microenvironment. The targeting basis of the CAQK peptide lies in this elevated expression of CSPGs in the injured region. Although PLX5622 nanoparticles may also be taken up by other neural cells, the pharmacological mechanism of PLX5622 is through inhibition of the CSF-1R signaling pathway, which is predominantly expressed in microglia within the central nervous system. Therefore, even if the nanoparticles are internalized by other cell types, their primary biological effect remains confined to microglia and would not significantly affect other neural cells.

These nanoparticles were further embedded within an ROS-responsive antioxidant hydrogel (GelMA-PPS). GelMA-PPS/P was delivered directly to the injured brain tissue via intraoperative local injection, enabling site-specific administration without relying on systemic passage across the BBB. As a biodegradable material, the GelMA hydrogel gradually degrades in vivo. Notably, the introduction of PPS moieties in GelMA-PPS/P allows the hydrogel to respond to elevated ROS levels in the post-injury microenvironment, accelerating hydrogel degradation under oxidative conditions and further promoting the release of nanoparticles. In addition, PPS itself scavenges excess ROS, helping to improve local oxidative stress. During this process, the released PLX5622 nanoparticles, which are surface-modified with CAQK peptides, specifically bind to CSPGs that are highly expressed in the injured brain, enhancing their accumulation and retention at the trauma site and thereby achieving targeted delivery. These findings indicate that GelMA-PPS/P exhibits effective enrichment at the injury site, primarily owing to the advantage of localized drug delivery in achieving direct therapeutic targeting. Although intracranial administration has not yet been widely adopted in routine clinical practice, it presents a feasible approach for specific indications such as severe TBI. For example, intraoperative local delivery-such as direct application of therapeutic materials during decompressive craniectomy-represents a clinically relevant strategy. Therefore, this localized delivery approach combining hydrogels with nanomedicine offers insights for developing more precise targeted therapies in the future. Of note, in this study, the localization of nanoparticles within brain tissue was determined based on DAPI staining and the intrinsic fluorescence of preloaded markers. Consequently, conclusions regarding the spatial distribution of nanoparticles were primarily derived from morphological observation and anatomical landmarks, which may not fully capture the actual penetration and diffusion range of nanoparticles. Future studies will incorporate histological co-staining and three-dimensional imaging techniques to achieve a more accurate and quantitative assessment of nanoparticle-hydrogel distribution in brain tissue.

Although intracranial administration has not yet been widely integrated into routine clinical practice, it presents a feasible approach for specific indications such as severe TBI. For instance, intraoperative local delivery-such as direct application of therapeutic materials during decompressive craniectomy-represents a clinically viable strategy. Therefore, the localized delivery strategy combining hydrogels with nanomedicine also provides insights for developing more precise targeted therapies in the future. This system enables site-specific microglial depletion and effective modulation of neuroinflammation, thereby reducing oxidative stress and exerting neuroprotective effects.

Activation of neuroinflammatory responses following TBI induces excessive production of ROS, while the accumulation of ROS further amplifies inflammation, forming a pathological positive feedback loop that exacerbates brain tissue damage [11]. Therefore, disrupting this vicious cycle is critical for mitigating secondary brain injury. In this study, we developed a GelMA-PPS/P hydrogel that exerts neuroprotective effects through a dual mechanism: initially scavenging excessive ROS via ROS-responsive PPS moieties, followed by the targeted delivery of PLX5622 nanoparticles-facilitated by CAQK peptide modification-to selectively deplete microglia and suppress neuroinflammation. RNA sequencing analysis revealed that genes associated with inflammatory activation and ROS-induced damage were significantly downregulated in the ipsilateral cortex of GelMA-PPS/P-treated TBI mice. These findings provide transcriptomic evidence supporting the efficacy of the GelMA-PPS/P in attenuating neuroinflammation and oxidative stress following TBI.

Dynamic monitoring and evaluation of intracerebral neuroinflammation following GelMA-PPS/P treatment is crucial for assessing its therapeutic efficacy in TBI models. Given that excessive or sustained microglial activation post-TBI may trigger a neuroinflammatory cascade-or “cytokine storm”-and exacerbate secondary brain injury, the ability to monitor this process in vivo is of significant importance. Microglia, as key regulators of neuroinflammation, participate in the inflammatory response through diverse molecular and signaling pathways and are thus considered critical indicators for inflammation assessment. In this study, we employed an innovative approach using the PET tracer ¹⁸F-DPA-714 combined with small-animal PET-CT imaging to perform longitudinal [54], in vivo monitoring of neuroinflammatory dynamics in TBI mice treated with GelMA-PPS/P. This was complemented by immunohistochemical analysis of IBA-1, a specific marker of activated microglia, and ELISA-based quantification of major proinflammatory cytokines to comprehensively evaluate the anti-inflammatory effects of GelMA-PPS/P. Following treatment, ¹⁸F-DPA-714 PET-CT imaging revealed reduced tracer uptake, IBA-1 immunoreactivity was markedly suppressed, and proinflammatory cytokine levels were significantly decreased, collectively indicating effective microglial depletion and attenuation of neuroinflammation. In addition, both in vitro and in vivo experiments demonstrated that GelMA-PPS/P also reduced ROS levels and alleviated oxidative stress following TBI, highlighting its dual anti-inflammatory and antioxidant properties.

Following TBI, the accumulation of neuroinflammation and ROS creates a highly cytotoxic microenvironment, contributing to extensive neuronal damage [20]. The GelMA-PPS/P system effectively mitigates these initiating pathological factors by alleviating neuroinflammatory responses and reducing oxidative stress, thereby protecting neuronal integrity and minimizing tissue loss. In parallel, disruption of the BBB and the subsequent development of brain edema further exacerbate secondary brain injury. Brain edema remains a major clinical challenge in TBI management and is closely linked to BBB breakdown. TBI can directly compromise BBB structure, while excessive oxidative stress and inflammation further increase its permeability, allowing infiltration of fluids and inflammatory mediators into brain tissue, thus aggravating edema and establishing a detrimental feedback loop [55]. Therefore, targeting oxidative stress and inflammation to preserve BBB integrity is essential for the effective treatment of post-traumatic cerebral edema. In this context, GelMA-PPS/P exerts protective effects by attenuating inflammation and oxidative damage, thereby maintaining BBB function and reducing brain edema.

Due to the highly heterogeneous nature of TBI, the extent and progression of oxidative stress and inflammatory responses vary significantly between patients, emphasizing the need for precision medicine approaches. Clinical interventions must be tailored based on injury characteristics, severity, and individual patient profiles. In this context, our GelMA-PPS/P platform introduces a novel strategy with potential for targeted therapeutic application. Nonetheless, further research is required to define optimal treatment timing and develop adaptable intervention protocols. The interplay between PLX5622 and PPS in regulating redox imbalance, neuroimmune responses, and functional recovery remains insufficiently characterized, highlighting the necessity for deeper mechanistic insights. Additionally, continued refinement of the material system is critical to ensure clinical compatibility. Meanwhile, in the current study we used only male mice to maintain experimental control and due to resource limitations in this preliminary research phase, which precluded concurrent investigations in female animals. Future studies will include validation in female subjects to comprehensively evaluate the applicability and efficacy of this therapeutic strategy across sexes. Additionally, the targeting ability of CAQK peptides was demonstrated specifically in the context of brain injury, where CSPGs are locally upregulated following BBB disruption and neuroinflammation. Although CSPGs can also be elevated in other injured tissues, our findings are limited to brain injury, and further research is needed to assess tissue selectivity. Moreover, we fully acknowledge the limitations associated with 405 nm light-mediated crosslinking. To enhance the clinical feasibility of our hydrogel system, future work will explore the use of lower-energy visible light sources or the development of novel injectable hydrogels that do not require light-crosslinking, thereby improving safety and operability in clinical settings. Recognizing the current study’s limitations, future efforts should focus on expanding translational potential through comprehensive optimization and validation.

Conclusion

In this study, we designed for the first time a novel nanoparticle-hydrogel system with the dual ability to selectively deplete microglia at the injury site and provide antioxidant effects. After injection into the injured area, GelMA-PPS/P self-assembled into a gel under 405 nm blue light, adhering to the injury site. This system responds to and scavenges ROS while releasing encapsulated PLX5622 nanoparticles. The surface of these PLX5622 nanoparticles is functionalized with CAQK peptides, which specifically bind to CSPGs that are highly expressed in the injured area. Both in vitro and in vivo experiments demonstrated that GelMA-PPS/P can reduce microglia, suppresses neuroinflammation, and mitigates excessive oxidative stress after TBI. Furthermore, GelMA-PPS/P preserved BBB integrity and reduced brain edema, ultimately improving neurological deficits following TBI. Therefore, this study provides a new therapeutic strategy for the treatment of TBI.

Author contributions

J. F. conceived and designed this study. Y. H., J.G., M. X., Y. Ma. and W. W. performed experiments and/or analyzed data and/or prepared the figures. Y. H., Q. F., Z. H. and W. Q. analyzed data. Y. H. and Q. M. wrote the first version of the manuscript. J. F., Q. M. and J. G. revised the manuscript. J. F., J. J., W. W. and Y.H. funded this research. All authors have read and approved the final manuscript.

Funding

This study was supported by grants from the National Natural Science Foundation of China (82071358, 82371379, 82372501 and 82401610), the Program of Shanghai Academic Research Leader (21XD1422400), Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Support (02.101005.001.30.30 A), the Project of Shanghai Medical And Health Development Foundation (20224Z0012) and the PhD Student Innovation Cultivation Fund of Shanghai Jiao Tong University School of Medicine (24KCPYYB009).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

All animal experiments complied with the ARRIVE guidelines and were carried out in accordance with the National Research Council’s Guide for the Care and Use of Laboratory Animals. All animal experiments in this research were approved by the Ethical Review Board of Renji Hospital (RJ-2022-0820). CSF samples from TBI patients were acquired from Ren Ji Hospital, Shanghai Jiao Tong University School of Medicine following approval by the institutional research ethics board (No. LY2023-018-B).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.