Magnesium silicate nanosheets enable sustained hydrogen release to attenuate secondary brain injury following intracerebral hemorrhage

Abstract

Background

Intracerebral hemorrhage (ICH) is a devastating stroke subtype with high mortality and disability, primarily driven by secondary brain injury involving oxidative stress, neuroinflammation, and apoptosis. Molecular hydrogen (H₂) exhibits potent neuroprotective effects; however, its clinical translation is limited by poor bioavailability and the lack of sustained delivery strategies.

Methods

pH-responsive magnesium silicate nanosheets (MGNs) were synthesized and characterized for morphology, composition, and hydrogen-release behavior. A collagenase-induced ICH mouse model was used to evaluate the therapeutic efficacy of orally administered MGNs. Neurological outcomes, brain edema, histopathology, apoptosis, and inflammatory responses were assessed using behavioral tests, staining, Western blotting, and ELISA. Therapeutic performance was compared with conventional 3% hydrogen inhalation, and biosafety was systematically evaluated.

Results

MGNs exhibited a two-dimensional structure and enabled sustained hydrogen release, particularly under acidic conditions. MGNs treatment significantly improved neurological function, reduced brain edema and hematoma volume, preserved neuronal integrity, and attenuated secondary brain injury in a dose-dependent manner. Mechanistically, MGNs suppressed neuronal apoptosis and reduced pro-inflammatory cytokine levels. MGNs demonstrated superior neuroprotection compared with hydrogen inhalation and showed favorable biosafety profiles.

Conclusions

MGNs represent a safe and effective hydrogen-releasing nanoplatform that alleviates secondary brain injury after ICH, highlighting their potential for translational neuroprotective therapy.

Graphical Abstract

Introduction

Intracerebral hemorrhage (ICH) is a life-threatening cerebrovascular disorder that accounts for 10%–25% of all strokes [1]. It remains a major cause of death and disability worldwide. Despite improvements in medical care, the mortality rate of ICH remains high, reaching 30%–40%. Over half of the survivors experience long-term neurological deficits [2–4]. The pathophysiology of ICH consists of a primary phase and a secondary phase [5]. The primary phase involves hematoma formation due to vessel rupture. The secondary phase is characterized by oxidative stress, neuroinflammation, mitochondrial dysfunction, blood–brain barrier (BBB) disruption and apoptosis [6]. These pathological processes lead to progressive neuronal injury and poor clinical outcomes. Current clinical strategies, including blood pressure control and surgical hematoma evacuation, mainly target the primary phase [7–9]. However, large clinical trials such as INTERACT and MISTIE have failed to improve long-term prognosis [10, 11]. This highlights the need for therapies that can effectively mitigate secondary brain injury.

Hydrogen gas (H₂) has emerged as a promising neuroprotective agent [12]. It shows selective antioxidants, anti-inflammatory, and anti-apoptotic effects [13]. H₂ scavenges highly reactive radicals (•OH, ONOO⁻) [14], activates endogenous antioxidant defenses via the Nrf2/HO‑1 pathway, suppresses pro‑inflammatory cytokines and signaling cascades, regulates apoptosis and autophagy, preserves mitochondrial function, and modulates immune cell activity to maintain homeostasis [15]. In ICH models, H₂ inhalation reduces oxidative damage, suppresses apoptosis, and improves neurological function [16, 17]. However, H₂ has low solubility in biological fluids and a short half-life. Traditional delivery methods cannot achieve targeted or sustained brain exposure. This limits its therapeutic effect [18]. Therefore, an efficient delivery system is required to overcome these limitations.

Nanomaterials offer great potential in biomedical applications. Their tunable properties and surface chemistry enable targeted and responsive delivery. Two-dimensional nanosheets such as layered double hydroxides (LDHs) and graphene oxide (GO) have been explored for therapeutic use [19–21]. Their high surface area, biocompatibility, and modifiability make them ideal carriers. Magnesium silicate nanosheets (MGNs) are a new class of 2D nanomaterials derived from natural magnesium–silicate minerals such as talc and forsterite. The silicon and magnesium components form a layered silicate framework with abundant surface-active sites [22]. The Si–O tetrahedral structure confers mechanical stability and pH responsiveness. In acidic environments, such as the stomach (pH 1.5–3.5), the Si–O bonds in MGNs undergo hydrolysis. This reaction leads to the continuous release of hydrogen gas [23]. Due to its small molecular size, H₂ diffuses rapidly through tissues and reaches brain lesions via systemic circulation. This enables targeted intervention in secondary injury after ICH. MGNs also show excellent biocompatibility and low immunogenicity. Their surface can be modified to improve targeting or drug loading [23].

Based on these properties, magnesium silicate nanosheets are particularly well suited to address the unmet need for a safe, sustained, and non-invasive hydrogen delivery strategy in intracerebral hemorrhage. Their ability to generate hydrogen in response to acidic environments provides a strong rationale for exploring their therapeutic potential in attenuating secondary brain injury following ICH. Therefore, this study aims to develop pH-responsive magnesium silicate nanosheets as a hydrogen delivery platform and explore their neuroprotective effect in ICH. We synthesized and characterized MGNs and evaluated their therapeutic role in a mouse model of ICH. We hypothesized that MGNs release H₂ in the stomach, which is absorbed into the systemic circulation and subsequently crosses the blood–brain barrier to reach brain lesions. By reducing oxidative stress, inflammation, and apoptosis, MGNs could attenuate secondary brain injury. This work provides a new strategy for ICH treatment and advances the application of nanomaterials in neurological diseases.

Materials and methods

Materials

Magnesium silicide(Mg₂Si) powder with an average particle size of 5 μm was purchased from Shanghai Xiangtian Nanomaterials Co., Ltd. (Shanghai, China) and used as received without further purification. The surface morphology of the samples was characterized using a field-emission scanning electron microscope (SEM, Gemini Sigma 300, ZEISS, Germany) equipped with an energy-dispersive X-ray spectroscopy system (EDS, Xplore 30, Oxford Instruments, UK) for elemental analysis. Atomic force microscopy (AFM) measurements were conducted on a Dimension ICON system (Bruker, USA). X-ray diffraction (XRD) patterns were collected using a D8 Advance diffractometer (Bruker, Germany) to analyze the crystalline structure. The hydrogen concentration was monitored using a microsensor (H₂ Microsensor, Unisense, Denmark).

Preparation and characterization of MGNs

Multilayer magnesium silicate nanosheets were synthesized and characterized at the Hydrogen Science Research Center, Shanghai Jiao Tong University [24]. The nanosheets were prepared by ball-milling Mg₂Si into microparticles, followed by ultrasonic crushing to achieve a two-dimensional nanosheet structure. The morphology and size of MGNs were observed using SEM and AFM. Elemental composition was confirmed by EDS, while crystal structure was analyzed using XRD. The hydrogen release capacity of MGNs was assessed using a hydrogen meter in different pH environments, demonstrating sustained hydrogen release, particularly under acidic conditions.

Animals and treatment

The Animal Ethics Committee of Shandong Second Medical University approved all experimental protocols (2024SDL659). All animal experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals (NIH Guide). Male C57BL/6J mice, 8–10 weeks of age, were kept under SPF standard housing conditions with free access to food and water. The ICH model was established by collagenase injection. Thirty minutes after induction, MGNs were administered to randomly assigned groups of mice. To meet the medication timeliness for treating cerebral hemorrhage, we set up concentration gradients of 0,75,125,175 mg/kg for oral gavage after ICH to find the optimal therapeutic dose. Control mice received an equivalent volume of normal saline.

Intracerebral hemorrhage model

ICH was induced by stereotaxic injection of collagenase VII-S into the right striatum of anesthetized mice. Post-surgery, mice received postoperative care including analgesia with 5% lidocaine. Animals were euthanized 72 h after ICH for further analysis.

Hematoxylin-eosin (HE) staining

Mice were euthanized 72 h after ICH induction, and brains were harvested, fixed in 4% paraformaldehyde at 4 °C for 48 h, dehydrated in graded ethanol, cleared with xylene, and embedded in paraffin. Coronal Sect. (5 μm) were obtained, deparaffinized, rehydrated, and stained with hematoxylin (5 min), followed by rinsing, differentiation in 1% hydrochloric acid–ethanol, and counterstaining with eosin Y (3 min). After subsequent dehydration and xylene treatment, sections were coverslipped and analyzed under a 400× microscope to assess neuronal structure and inflammation.

Nissl staining

Adjacent 5-µm paraffin sections were deparaffinized and rehydrated. Sections were stained with 0.1% toluidine blue (pH 4.0) at 37 °C for 15 min, differentiated in 70% ethanol, washed, dehydrated, cleared, coverslipped, and observed at 400×. Nissl bodies in the cortex and striatum were counted per HPF in the peri-hematomal region using ImageJ; 5 HPFs/section and 3 sections/mouse were analyzed.

Transmission electron microscope (TEM)

72 h after ICH, brain tissues surrounding the hemorrhage site were collected and fixed in 2.5% glutaraldehyde. Samples were dehydrated, embedded, and sectioned for TEM analysis. Mitochondrial morphology was assessed using ImageJ software.

Modified neurological severity score (mNSS) assessment​

72 h after ICH, mNSS was performed by two independent investigators blinded to the experimental groups. The mNSS is a composite score that evaluates motor, sensory, reflex, and balance functions in mice. Motor function was assessed by observing spontaneous activity, limb symmetry during movement, and resistance to lateral push. Sensory function was tested using vibrissae touch response, visual placing, and proprioceptive placing. Reflexes were evaluated by assessing the righting reflex, corneal reflex, and pinna reflex. Balance was determined by the ability of the mouse to remain on a narrow beam. Each category was scored from 0 (normal function) to a maximum score depending on the deficit, with a total maximum score of 18 points. Higher scores indicate more severe neurological deficits.

Brain edema measurement​

At 72 h following ICH induction, mice were euthanized and their brains quickly extracted. The cerebellum and brainstem were removed, and the remaining cerebral hemispheres were immediately weighed to determine the wet weight (WW). Samples were then oven-dried at 105 °C for 24 h until a constant dry weight (DW) was achieved. Total brain water content was calculated as: [(WW − DW) / WW] × 100%.For regional edema assessment, brains were sectioned coronally into 2 mm slices using a brain matrix. The peri-hematomal tissue and the corresponding contralateral region were dissected, weighed for WW, dried, and reweighed for DW, from which the regional water content was determined using the same formula.

Tunel staining​

Section (5 μm) from paraffin-embedded brain samples were deparaffinized, rehydrated, and incubated with proteinase K (20 µg/mL in Tris–HCl, pH 7.4) at 37 °C for 15 min to permeabilize cells. After PBS rinsing, endogenous peroxidase activity was blocked by treating the sections with 3% hydrogen peroxide in methanol for 10 min. The TUNEL reaction mixture (Roche, Mannheim, Germany) was then applied for 60 min at 37 °C in a humidified chamber according to the manufacturer’s protocol. Following another PBS wash, nuclei were counterstained with DAPI (4′,6-diamidino-2-phenylindole) for 5 min. Apoptotic cells were identified as TUNEL-positive under a Leica DMi8 fluorescence microscope (400×). Counts were performed in at least five high-power fields in the peri-hematomal zone per section, with three sections evaluated for each mouse.

Western blot analysis​

At 72 h post-ICH, mice were euthanized, and the peri-hematomal brain tissue was carefully isolated. Samples were homogenized in radioimmunoprecipitation assay (RIPA) buffer (Beyotime, China) supplemented with protease and phosphatase inhibitors. The homogenates were centrifuged at 12,000 × g for 15 min at 4 °C, and the resulting supernatants were collected as total protein extracts. Protein concentrations were determined using a bicinchoninic acid (BCA) assay kit (Thermo Fisher Scientific, USA). Equal amounts of protein (30–50 µg) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and transferred onto polyvinylidene fluoride (PVDF) membranes (Abcam, USA). Membranes were blocked with 5% non-fat milk in Tris-buffered saline containing 0.1% Tween-20 (TBST) for 1 h at room temperature, followed by overnight incubation at 4 °C with primary antibodies diluted in 5% bovine serum albumin (BSA) in TBST. The primary antibodies used for protein included anti-Bax (Proteintech, Cat. No. 50599-2-Ig; 1:2000), anti-Bcl-2 (Proteintech; Cat. No. 12789-1-AP; 1:1000), anti-cleaved caspase-3 (Proteintech; Cat. No. 25128-1-AP; 1:500), anti-IL-1β (Proteintech; Cat. No.16806-1-AP; 1:500), anti-IL-6 (Proteintech; Cat. No. 21865-1-AP; 1:1000), anti-IL-18 (Proteintech; Cat. No. 10663-1-AP; 1:1000), and anti-TNF-α (Proteintech; Cat. No. 17590-1-AP; 1:1000). After TBST washes, membranes were incubated with horseradish peroxidase–conjugated secondary antibodies (Proteintech; Cat. SA00001-2 1:5000) for 1 h at room temperature. Protein bands were detected using an enhanced chemiluminescence (ECL) kit (Thermo Fisher Scientific, USA) and visualized with a digital chemiluminescence imaging system. Band intensities were analyzed using ImageJ software, and protein levels were normalized to Tubulin (Proteintech; Cat. No. 66031-1-Ig 1:8000).

Enzyme-linked immunosorbent assay (ELISA) kit (Solarbio) analysis for IL-1β, IL–6, IL-18, and TNF–α

At 72 h post-ICH, peri-hematomal brain tissue was excised, homogenized in ice-cold RIPA buffer supplemented with protease inhibitors, and centrifuged at 12,000 × g for 15 min at 4 °C. The resulting supernatants were collected for cytokine quantification. ELISA kits (Solarbio, China) were used following the manufacturer’s instructions. For standard curve preparation, serial dilutions of the supplied standards were plated (50 µL/well). For each sample, 40 µL of dilution buffer and 10 µL of tissue supernatant were added to the wells. Subsequently, 100 µL of enzyme-conjugated antibody was dispensed into all wells except the blanks. Plates were sealed and incubated at 37 °C for 1 h before further processing in accordance with kit protocols.

Statistical analysis

All values are expressed as the mean ± standard deviation (SD). Each experiment was independently repeated at least three times to ensure reproducibility. Data analyses were conducted using GraphPad Prism 9.0 software (GraphPad Inc., San Diego, CA, USA). For multiple group comparisons (≥ 3 groups), one-way analysis of variance (ANOVA) was applied, followed by Tukey’s post hoc test. Statistical significance was indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. A p-value < 0.05 was considered statistically significant, while n.s. indicated no significant difference.

Results

Characterization of MGNs

We selected magnesium silicide (Mg₂Si) as the starting material for its known ability to release hydrogen under certain conditions. Initially, we dispersed Mg₂Si into smaller microparticles through a ball milling process. Following this, we used ultrasonic crushing and centrifugation to obtain a two-dimensional nanosheet structure (Fig. 1A, E). The SEM images confirmed that these nanosheets have a sheet-like structure with a length of approximately 7 μm.

Fig. 1.

MGNs construction process. (A) Schematic diagram illustrating the preparation process of MGNs. (B) SEM image showed the sheet-like structure of MGNs (C) Elemental mapping via EDS confirms the presence of magnesium (Mg) and silicon (Si) in MGNs. (D) XRD pattern displaying the crystal structure of MGNs with prominent peaks at 24.3°, 40.1°, and 72.9°, corresponding to Mg₂Si. (E) Macroscopic characterization of MGNs. All experiments were performed in triplicate (n = 3)

To verify the composition of these nanosheets, EDS was performed, which showed that the nanosheets contain both magnesium (Mg) and silicon (Si) elements (Fig. 1B, C). We analyzed the crystalline structure using powder XRD, which revealed three prominent peaks at 24.3°, 40.1°, and 72.9°, corresponding to specific planes of Mg₂Si. These results indicate that the exfoliation process used to create the nanosheets did not alter the original crystalline phase of the material (Fig. 1D).

.

pH-Responsive hydrogen release and morphological characterization of MGNs

Next, we tested the ability of these nanosheets to release hydrogen continuously under different pH conditions using an electrode method. The results demonstrated that MGNs could sustain hydrogen release for more than 30 h. However, the rate of hydrogen release varied significantly depending on the pH of the surrounding environment. In an acidic solution (pH 5.0), the hydrogen release was the fastest, reaching its peak within about 5 min and then gradually decreasing. In contrast, in an alkaline solution (pH 9.0), it took nearly 18 min to reach the peak release rate (Fig. 2A). AFM provided additional confirmation of the nanosheets’ dimensions, showing that while the sheets are approximately 1 μm in diameter, they are only 23 nm thick, confirming their ultra-thin structure (Fig. 2B, C). After 30 h, the total hydrogen concentration in the acidic solution reached approximately 2500 µmol/L (Fig. 2A, D). This suggests that MGNs are particularly effective at producing hydrogen rapidly in acidic environments, which could be highly beneficial for emergency treatment of acute brain injuries where immediate intervention is critical.

Fig. 2.

Analysis and identification of MGNs. (A) Hydrogen release profile of MGNs in phosphate-buffered saline (PBS) at pH 5.0, 7.4, and 9.0, showing sustained release over 36 h. (B) AFM image depicting the ultra-thin structure of MGNs. (C) Dimension distribution of MGNs as measured by AFM. (D) Total hydrogen concentration in the reaction system over 48 h, showing rapid release in acidic conditions. All experiments were performed in triplicate (n = 3)

Neuroprotective effects of MGNs at different concentrations in ICH mice

To assess the neuroprotective effect of MGNs following ICH, we evaluated motor and behavioral performance using the mNSS. Mice treated with MGNs showed significantly improved neurological scores compared to the control group (Fig. 3A). These results suggest that MGNs effectively preserve neurological function after ICH.Brain edema is a common pathological consequence of ICH. To further evaluate the protective effect of MGNs, brain water content was measured. Mice in the MGN-treated groups exhibited significantly reduced brain water content compared to controls (Fig. 3B). This indicates that MGNs can attenuate cerebral edema, possibly by preserving the blood–brain barrier or reducing cell death. Histopathological analysis provided further insight into the protective mechanism. HE staining revealed severe neuronal damage in the perihematomal area of ICH mice, including nuclear pyknosis, eosinophilic cytoplasm, and infiltration of inflammatory cells (Fig. 3C).

Fig. 3.

Protective effects of MGNs at different concentrations against intracerebral hemorrhage. (A) Neurological function assessed by mNSS shows significant improvement in MGNs-treated mice (n = 10). (B) Brain water content measurement indicating decreased edema in MGNs-treated mice. (C-D) HE and Nissl staining demonstrate less cellular damage and more Nissl bodies treated with 75, 125, 175 mg/kg concentrations of MGNs (n = 3). All data are presented as mean ± SD, analyzed using one-way ANOVA with Tukey’s post-hoc test (***p < 0.001)

In contrast, mice treated with 175 mg/kg MGNs showed improved neuronal morphology, reduced intercellular space, and decreased inflammatory infiltration. Nissl staining (Fig. 3D) demonstrated a loss of Nissl bodies in neurons from the ICH group compared to the sham group. Mice treated with MGNs exhibited a higher number of preserved Nissl bodies. Since Nissl bodies serve as markers of neuronal integrity and function, this suggests that MGNs reduce brain tissue injury. The neuroprotective effect displayed a clear dose-dependent trend.

MGNs inhibit apoptosis following intracerebral hemorrhage

To evaluate the neuroprotective and anti-apoptotic potential of MGNs following ICH, TUNEL staining was performed to detect neuronal apoptosis in mouse brain sections. Mice received MGNs at doses of 75, 125, or 175 mg/kg. Abundant TUNEL-positive cells (red) were localized mainly in the peri-hematomal area of untreated ICH brains, whereas MGNs administration markedly reduced their number (Fig. 4A). The reduction in apoptotic cells exhibited a clear dose-dependent pattern, with higher MGNs doses producing greater protective effects. Quantitative analysis of TUNEL-positive cells reinforced this trend, indicating consistent neuroprotection across treatment groups (Fig. 4B).

Fig. 4.

Concentration-dependent inhibition of post-ICH apoptosis by MGNs. (A-B) TUNEL staining (red) comparing neuronal death levels around the lesion area in stroke mouse brain tissues treated with 75, 125, 175 mg/kg concentrations of MGNs, with nuclei stained with DAPI (blue), scale bar: 50µ. The percentage of apoptotic cells in the total cell count. (C-F) Representative Western blot images and corresponding quantitative analyses of Bax, Bcl‑2, and cleaved caspase‑3 in brain tissues from Sham, ICH, and ICH + MGNs groups (n = 3). Data are expressed as mean ± SD and were evaluated by one-way ANOVA (*p < 0.05, **p < 0.01, ***p < 0.001). n.s. indicates no statistically significant difference

To investigate potential mechanisms, expression levels of apoptosis-associated proteins were examined via Western blot. Targets included Bax, Bcl‑2, and cleaved caspase‑3 (Fig. 4C). In the ICH group, Bax and cleaved caspase‑3 were significantly upregulated, whereas Bcl‑2 expression was suppressed, suggesting pronounced apoptosis. MGNs treatment reversed these alterations, decreasing Bax and cleaved caspase‑3 levels while enhancing Bcl‑2 expression (Fig. 4D–F). Collectively, these data support that MGNs mitigate neuronal apoptosis by modulating the Bax/Bcl‑2 balance and inhibiting caspase‑3 activation in a dose-dependent manner, underscoring their therapeutic potential for post‑ICH neuroprotection.

MGNs suppress neuroinflammation following intracerebral hemorrhage

ICH is frequently accompanied by a pronounced neuroinflammatory response. To assess the anti-inflammatory potential of MGNs, we examined the expression of representative pro-inflammatory cytokines, including IL‑1β, IL‑6, IL‑18, and TNF‑α. Both Western blot and ELISA analyses showed that MGNs treatment markedly decreased the levels of these cytokines compared with the ICH group. This suggests that MGNs attenuate post‑ICH inflammation by limiting the release of inflammatory mediators.

In the Western blot assays (Fig. 5A–D), protein levels of IL‑1β, IL‑6, IL‑18, and TNF‑α were substantially reduced in MGN‑treated mice relative to untreated ICH controls, indicating modulation of inflammation‑related molecular pathways. Consistent with these findings, ELISA measurements revealed significantly lower concentrations of these cytokines in the brain tissue of MGNs‑treated animals (Fig. 5E–H). Together, these results demonstrate that MGNs alleviate neuroinflammation following ICH by downregulating pro‑inflammatory cytokine production.

Fig. 5.

MGNs exert neuroprotective effects against ICH by suppressing post-hemorrhagic neuroinflammation. (A-D) Western blot and (E-H) ELISA analysis of pro-inflammatory cytokines (IL-1β, IL-6, IL-18, TNF-α) levels in brain tissues, showing significant reduction in the ICH + MGNs group compared to ICH controls. All data are presented as mean ± SD, analyzed using one-way ANOVA with Tukey’s post-hoc test (*p < 0.05, **p < 0.01, ***p < 0.001)

MGNs perform better than traditional hydrogen therapy

To compare the therapeutic effects of MGNs with traditional hydrogen therapy, we evaluated the efficacy of MGNs (175 mg/kg) versus inhaled 3% H₂ in mice after ICH. Both treatment methods were effective in reducing the hematoma volume after ICH (Fig. 6A, C). However, the reduction was significantly greater in the MGNs-treated group compared to the group treated with 3% H₂, indicating that MGNs have a superior therapeutic effect. Western blot analysis and TUNEL staining showed that MGNs were more effective in reducing the levels of apoptotic proteins, such as Bax and cleaved-caspase-3, while increasing the level of the anti-apoptotic protein Bcl-2 (Fig. 6B, D-H). This suggests that MGNs provide stronger protection against cell death compared to traditional hydrogen therapy.

Fig. 6.

The efficacy of MGNs is better than the conventional 3% H₂ inhalation method. (A, C) Representative images and quantification of hemorrhage volume in the Sham, ICH, ICH + 3% H₂, and ICH + MGNs groups, showing greater reduction in hematoma volume with MGNs treatment (n = 10). (B-F) Western blot analysis of apoptotic markers (Bax, Bcl-2, cleaved-caspase-3) indicates superior protection against cell death by MGNs compared to 3% H₂. (G-H) TUNEL staining shows lower levels of apoptosis in the MGNs-treated group. All data are presented as mean ± SD, analyzed using one-way ANOVA with Tukey’s post-hoc test (*p < 0.05, **p < 0.01, ***p < 0.001)

Biosafety evaluation of MGNs in mouse model

To assess the biosafety of MGNs, histological analysis was performed using HE staining on major organs, including the heart, liver, spleen, stomach, and kidney. No signs of structural damage or inflammatory infiltration were observed in tissues from MGN-treated mice compared to controls. In particular, hepatocytes and renal tubular cells maintained normal morphology, with no apparent apoptosis or edema (Fig. 7A). These findings suggest that MGNs do not cause significant tissue toxicity within the tested dosage range.

Fig. 7.

MGNs demonstrate favorable biosafety profiles. (A) H&E staining and (B-E) biochemical assays (ALT, AST, UREA, CREA-S) reveal no significant toxicity or side effects in MGNs-treated mice compared to controls. All data are presented as mean ± SD, analyzed using one-way ANOVA with Tukey’s post-hoc test. n.s. stands for not significant

To further evaluate systemic biosafety, serum biochemical parameters were analyzed. These included alanine aminotransferases (ALT) and aspartate aminotransferase (AST) for liver function, as well as UREA and serum creatinine (CREA-S) for kidney function. No significant differences in ALT, AST, UREA, or CREA-S levels were found between MGN-treated mice and controls (Fig. 7B–E). These results indicate that MGNs do not impair hepatic or renal function, supporting their favorable biocompatibility and systemic safety.

MGNs were synthesized from Mg₂Si by ball milling followed by exfoliation. Upon oral administration, MGNs released hydrogen gas in response to the acidic gastric environment. They modulated apoptosis-related proteins (Bax, Bcl-2, and cleaved caspase-3) and inflammatory cytokines such as IL-1β. These actions exerted neuroprotective, anti-apoptotic, and anti-inflammatory effects. As a result, secondary brain injury following ICH was alleviated.

Discussion

ICH remains a catastrophic neurological disease, with secondary brain injury driven by oxidative stress, inflammation, and apoptosis being the primary cause of poor outcomes. Despite decades of research, no pharmacological therapy has successfully translated to the clinic, highlighting the need for strategies that can intervene effectively in the secondary injury cascade [25].

Hydrogen therapy has long been recognized as a potential neuroprotective strategy. Early studies established that low-concentration hydrogen inhalation reduces oxidative damage and improves neurological recovery after experimental ICH [16, 26–28]. However, the protective effects are short-lived, while the pathophysiological cascade persists for nearly a week. This temporal limitation, together with the reliance on inhalation devices, has severely restricted clinical translation [29, 30]. Our study builds directly on this foundation by addressing the unmet need for a delivery platform capable of sustaining hydrogen availability during the entire secondary injury period (Fig. 8).

Fig. 8.

Schematic illustration of the preparation, hydrogen release, and neuroprotective mechanism of MGNs in ICH

In this context, we developed MGNs, which represent a conceptual advance over previous hydrogen delivery systems. Unlike inhalation, MGNs exploit gastric acidity to generate hydrogen in situ, thereby achieving both rapid release and prolonged systemic exposure [18, 31, 32]. This distinguishes MGNs from localized pH-responsive nanoplatforms such as ammonia borane–loaded polydopamine nanoparticles, which are confined to restricted tissue environments and require invasive delivery [33]. By contrast, systemic hydrogen distribution achieved by MGNs is particularly advantageous in ICH, where injury is diffuse and extends beyond immediate hematoma.

Compared with other nanotherapeutic strategies, MGNs offer a broader mechanistic profile. For example, revascularization strategies have employed nanocarriers such as TPA-DPNs and Au@MSNs to protect thrombolytic agents (e.g., tPA and uPA) and enable stimulus-responsive release, thereby prolonging drug activity, extending the therapeutic window, and reducing the risk of hemorrhage [34]. Reactive oxygen species (ROS) clearance strategies utilize platforms such as CeO₂@ZIF-8, aC5a-FNA, and POM nanoclusters to eliminate ROS and reactive nitrogen species (RNS) through enzyme-mimetic catalytic activity or targeted binding, thereby alleviating oxidative stress-induced cerebral injury [35]. Anti-inflammatory, including PEG/cRGD liposomes loaded with 9-AA [36], polymeric micelles encapsulating rapamycin [37], and PTNPs [38], have been designed to modulate inflammatory signaling pathways, reprogram immune cell phenotypes, or block neutrophil infiltration, thus mitigating inflammatory cascades. In addition, neuroregeneration strategies employ biodegradable nanomaterials carrying catalase (CAT) and superoxide dismutase (SOD), MFIONs, or MNV to optimize the neuroregenerative microenvironment, enhance the targeting of reparative cells, and promote both anti-inflammatory effects and angiogenesis, ultimately facilitating neuronal regeneration and functional recovery [39]. Our findings suggest that MGNs act more directly and comprehensively: molecular hydrogen, owing to its small size and high diffusivity, crosses the blood–brain barrier and simultaneously modulates inflammatory and apoptotic pathways. This dual-targeting effect—evidenced by reduced cytokine release and inhibition of caspase-dependent apoptosis—sets MGNs apart from single-pathway nanotherapies, which have consistently shown incomplete protection in treatment.

The superiority of MGNs becomes more apparent when contrasted with existing pharmacological approaches. Nimodipine, for example, while effective in alleviating vasospasm after subarachnoid hemorrhage [40], failed to improve cognitive outcomes or survival in ICH patients in multicenter trials, reflecting its narrow, vessel-specific mechanism that does not address the oxidative and inflammatory cascades [41]. Similarly, in acute ischemic stroke (AIS), numerous neuroprotective agents have yielded no clinical benefit, underscoring the urgent need for new therapeutic strategies [42]. Edaravone, a potent free radical scavenger recommended by Chinese and Japanese stroke care guidelines, can neutralize hydroxyl radicals (·OH), nitric oxide radicals (NO·), and peroxynitrite anions (ONOO⁻), thereby alleviating cerebral edema and inhibiting delayed neuronal death [43]. Although Edaravone demonstrates efficacy in AIS, its pharmacokinetic limitations and the challenge of maintaining sustained antioxidant activity within brain parenchyma may restrict its impact in ICH [44]. In contrast, MGNs offer a distinct advantage by maintaining elevated hydrogen levels directly within brain tissue, enabling continuous neutralization of hydroxyl radicals and peroxynitrite—two pivotal mediators of ICH-induced oxidative stress. In this manner, our study advances the hydrogen-based therapeutic field from transient inhalation models to a more durable and clinically applicable strategy.

Biosafety remains an essential consideration for any nanomaterial intended for neurological use. Here, our data showing no hepatic or renal toxicity align with prior reports that magnesium-based materials exhibit good biocompatibility [23]. However, unlike conventional drugs, the long-term biodistribution and metabolic fate of MGNs require further study. By providing a detailed preclinical safety profile, our study contributes valuable reference data for the evaluation of nanomaterials in ICH therapy and beyond.

Several limitations of the present study should be acknowledged. First, our findings are derived from a murine ICH model, and validation in larger animal models with closer anatomical and physiological resemblance to humans will be essential to further support clinical translation. Second, although acute biosafety and toxicity were systematically evaluated, the long-term biodistribution, metabolic fate, and potential delayed effects of MGNs remain incompletely characterized and warrant further investigation.

Third, while MGNs demonstrated effective and sustained hydrogen release under acidic conditions in vitro, hydrogen release efficiency in vivo may vary due to individual differences in gastrointestinal pH. Notably, the pH value employed in this study (pH 5.0) is higher than that of the physiological gastric environment, which typically ranges from pH 1.5 to 3.5. Given that the hydrolysis of Si–O bonds is generally accelerated under more acidic conditions, hydrogen generation from MGNs in the stomach may occur at a faster initial rate than that observed in vitro, suggesting that our experimental conditions likely represent a conservative estimation rather than an overestimation of in vivo hydrogen release [45, 46]. Nevertheless, variability in gastric acidity across individuals and disease states may influence hydrogen release kinetics and systemic availability. Future studies should therefore systematically evaluate hydrogen generation profiles across a broader pH range and further optimize material composition and fabrication strategies to enhance hydrogen-loading capacity and achieve more precise and controllable release behavior.

In addition, the interactions between MGNs and host tissues require deeper exploration. Beyond their direct neuroprotective effects mediated by hydrogen release, MGNs may influence immune cell activation, endothelial function, and neurovascular coupling—processes that are increasingly recognized as critical contributors to secondary brain injury following ICH [25, 47, 48]. Moreover, MGNs are unlikely to function solely as a standalone therapy in clinical settings. Their combination with established interventions, such as surgical hematoma evacuation, may provide synergistic benefits by simultaneously alleviating primary mass effect and mitigating secondary injury processes, including oxidative stress, neuroinflammation, and blood–brain barrier disruption. A more comprehensive understanding of these interactions may also reveal novel therapeutic mechanisms and broaden the application of MGNs to other neurological disorders, such as ischemic stroke, traumatic brain injury, and neurodegenerative diseases, where oxidative stress and inflammation play central roles.

Conclusion

In summary, we developed MGNs as a hydrogen-releasing nanoplatform for ICH therapy. MGNs effectively reduced neuronal apoptosis and neuroinflammation, leading to improved neurological outcomes in ICH mice. Compared to conventional therapies, MGNs showed superior efficacy and favorable biosafety. These findings suggest the potential of MGNs in clinical translation and future application in neurological diseases.

Abbreviations

AFM

Atomic Force Microscopy

AIS

Acute Ischemic Stroke

ALT

Alanine Aminotransferase

AST

Aspartate Aminotransferase

BBB

Blood—Brain Barrier

CAT

Catalase

CREA

S—Serum Creatinine

EDS

Energy—Dispersive X—ray Spectroscopy

H₂

Hydrogen

HE

Hematoxylin—Eosin

ICH

Intracerebral Hemorrhage

IL

Interleukin

MGNs

Magnesium Silicate Nanosheets

mNSS

Modified Neurological Severity Score

ROS

Reactive Oxygen Species

SEM

Scanning Electron Microscopy

SOD

Superoxide Dismutase

TEM

Transmission Electron Microscopy

XRD

X—ray Diffraction

Author contributions

M.-T. H., Y.-S. W., and C.-S. M. conceived and designed the study. M.-T. H., A.-J. D., and Y.-S. W. were responsible for funding acquisition, methodology, project administration, resources, supervision, validation, and writing—review & editing. C.-S. M., B. H., and J.-R. G. performed data curation, formal analysis, visualization, and writing—original draft. L.-Y. Z., J.-F. G., and W.-J. Y. contributed to animal experiments, data collection, and discussion. C.-K. S. and Y.-X. L. conducted cell experiments and assisted with data analysis. All authors participated in manuscript revision, approved the final version, and agreed to be accountable for all aspects of the work.

Data availability

All data supporting the conclusions of this study can be obtained from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors state that they have no financial or personal conflicts of interest that could have influenced the research presented in this manuscript.