Bioactive manganese oxide nanoassembly as a highly sensitive magnetic resonance imaging probe for grading diagnosis of liver injury

Abstract

Real-time and accurate detection of drug-induced liver injury is critical for early intervention and treatment, yet clinically applicable visualization methods remain scarce. Based on the difference in glutathione (GSH) content between the normal liver and the livers with different-grade injury, we report a GSH-activated T₁-weighted magnetic resonance imaging (MRI) nanoprobe (C-MnO), which can be used for non-invasive real-time in vivo magnetic resonance imaging (MRI) of the consumption of GSH in the liver, thereby being applied for the graded diagnosis of liver injury. Due to the presence of abundant disulfide bonds in C-MnO, the T₁ MRI signal around it remained “quenched” until encountering GSH. When C-MnO enters the body and is efficiently absorbed by the liver, it will disassemble and degrade under the action of GSH, thereby activating the T₁ MRI signal. Therefore, this nanoprobe provides an effective visual method for capturing the changes in GSH content during different degrees of liver damages at an early stage, which is beneficial for the graded diagnosis and precision treatment of drug-induced liver injury. Comprehensive in vitro and in vivo studies demonstrate that C-MnO, as a GSH-activated T₁ MRI nanoprobe, enables real-time monitoring and graded diagnosis of drug-induced liver injury, effectively addressing current clinical limitations in the detection of liver injury.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04144-4.

Introduction

The liver is the central hub of various physiological processes and plays a core role in the metabolism of nutrients and the detoxification of exogenous substances [1–3]. Consequently, commonly used oral medications will be converted into various metabolites in the liver. Drug-induced liver injury represents one of the most prevalent metabolic liver disorders, typically resulting from excessive or prolonged medication exposure [4–6]. Currently, the primary clinical diagnostic approach for drug-induced liver injury relies on serological testing, which involves the detection of specific biomarkers in the blood, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP), etc [7, 8]. Clinical studies have demonstrated that abnormal levels of these biomarkers exhibit a positive correlation with drug-induced liver injury [4, 9]. However, serological testing presents several limitations: (1) it does not allow for in situ, real-time monitoring of disease progression; (2) its results can be confounded by other comorbidities; and (3) it often lacks sensitivity for early-stage liver injury assessment [10–12]. Liver biopsy remains the gold standard for diagnosing hepatic disorders, yet this method is highly invasive, prone to sampling errors, and carries potential complications [13]. Meanwhile, clinical imaging techniques—such as ultrasound (US) and computed tomography (CT)—can visualize structural abnormalities in liver tissue, aiding diagnosis. However, these modalities rely on macroscopic anatomical changes, which are non-specific, lack molecular-level insights, and often fail to distinguish between different disease stages with sufficient sensitivity [14, 15]. Given the inability of current methods to comprehensively characterize molecular alterations associated with liver injury, there is an urgent demand for the development of real-time, non-invasive, and clinically convenient detection strategies.

Currently, over 1,000 clinically used drugs are known to exhibit hepatotoxicity [16]. Isoniazid (INH), a first-line anti-tuberculosis agent, is among the most frequently implicated drugs in drug-induced liver injury [17–19]. The potential mechanisms of liver injury induced by INH include oxidative stress, mitochondrial dysfunction, endoplasmic reticulum stress, imbalance of bile acid transport and immune response [20, 21]. Studies have shown that the direct oxidation of drugs is the cause of covalent modification of proteins and cell damage [22]. Similarly, acetaminophen (APAP, paracetamol), a widely administered analgesic and antipyretic, is another common cause of drug-induced liver injury [23, 24]. However, excessive intake of APAP can lead to oxidative stress and nitrosation stress, which are overproduced through reactive oxygen species (ROS) and reactive nitrogen species (RNS). This will lead to the consumption of critical cellular antioxidants such as glutathione (GSH), triggering a cascade of signaling events that ultimately result in necrotic cell death [22]. As the most abundant non-protein biothiol in cells, GSH plays pivotal roles in maintaining redox homeostasis, facilitating xenobiotic metabolism, regulating intracellular signaling, and modulating gene expression. This important biothiol can protect cellular components from damage caused by free radicals and ROS while maintaining exogenous antioxidants in their reduced states [25–27]. GSH levels typically decline during liver injury, primarily due to the elevated generation of ROS or RNS that deplete GSH reserves [28]. Consequently, the glutathione concentration in liver tissue can serve as both a direct and vital biomarker for evaluating liver injury, and visual monitoring of glutathione levels in the liver is essential for early diagnosis and prognostic evaluation of liver injury [8]. Zheng et al. developed an innovative fluorescence-based detection system using thiol-modified gold clusters that release fluorescence upon GSH interaction, enabling in vivo liver injury monitoring [29, 30]. The fluorescence method for detecting liver injury has solved the challenges encountered in non-invasive monitoring of liver GSH depletion for a long time, opening up more possibilities for clinical applications. However, fluorescence imaging techniques face inherent limitations including limited tissue penetration depth, low spatial resolution, and autofluorescence interference [31–33]. In contrast, magnetic resonance imaging (MRI) has emerged as an important diagnostic tool for liver diseases due to its excellent spatial resolution and soft tissue contrast [34–36]. Clinically, in order to enhance the contrast between the lesion and normal tissues, magnetic resonance contrast agents are often used [37–40]. The principle of these agents mainly lies in the shortening of the T₁ or T₂ relaxation time of the surrounding protons by the paramagnetic centers, which then results in a distinct bright or dark signal in MRI, thereby improving the detection efficiency of the lesion [41–43]. In previous study, we synthesized amorphous FeₓMn_γO nanoparticles (AFMO-ZDS) as redox-responsive probes, enabling dynamic in vivo visualization of GSH-mediated biotransformation in liver via contrast-enhanced T₁-weighted magnetic resonance imaging (T₁ CE-MRI) [10]. This suggests that by constructing MRI probes with GSH response capabilities, the GSH levels in the liver can be effectively monitored using CE-MRI technology. This approach establishes a novel imaging strategy for the visual assessment of liver injury, offering potential clinical applications in hepatotoxicity evaluation and disease diagnosis.

Here, we have constructed a manganese oxide nano-assembly cross-linked by sulfhydrylated polyethylene glycol (labeled as C-MnO), and attempted to use it as a novel magnetic resonance nanoprobe for graded diagnosis of liver injury. This nano-assembly contains abundant disulfide bonds inside, so under the action of GSH, the disulfide bonds can break and disassemble, thereby achieving efficient activation of the T₁ MRI signal. As shown in Scheme 1, when C-MnO are injected into mice through the tail vein, most of it are taken up by the liver due to its large size. In the normal liver, the abundant presence of endogenous GSH initiates the breakage of disulfide bonds and the disassembly of C-MnO, releasing Mn²⁺ which is captured by GSH to form Mn²⁺-GSH chelates, thus accelerate the relaxation of surrounding water molecules and significantly activate T₁-weighted MRI signals [10]. In acute liver injury caused by INH or APAP, a slight decrease in GSH levels was observed due to the increase in ROS or RNS, resulting in the production of a relatively smaller amount of Mn²⁺-GSH chelates and a slightly lower T₁-weighted MRI signal. However, in cases of severe liver injury, due to the extensive hepatocellular damage, GSH is consumed in large quantities and cannot be replenished in time, resulting in its continuous maintenance at a low level. Consequently, the formation of Mn²⁺-GSH chelates is extremely limited, thereby making the T₁-weighted MRI signal almost impossible to be activated. Therefore, the C-MnO nanoassemblies can be used as an effective tool for real-time assessment of liver function, which is greatly conducive to the early detection and accurate evaluation of drug-induced graded liver injury. Notably, in response to the decline of GSH levels in damaged liver tissue—a key factor in maintaining intracellular redox homeostasis—the compensatory upregulation of glutathione reductase (GR) occurs, demonstrating higher sensitivity compared to conventional liver function markers. Furthermore, by integrating structural imaging with functional detection, this approach offers a multidimensional perspective for the precise grading diagnosis, mechanistic investigation, and therapeutic evaluation of liver injury, positioning it as a promising diagnostic tool in clinical practice.

Scheme 1.

Schematic illustration of drug-induced liver injury grading diagnosis using T₁ CE-MRI technology coupled with C-MnO nanoprobes

Results and discussion

Synthesis, characterization, and responsive performance of C-MnO

As shown in Fig. 1a, firstly, oleic acid-coated manganese oxide nanoparticles (labeled as MnO-OA) were synthesized using a high-temperature thermal decomposition method [44]. Then, meso-2,3-dimercaptosuccinic acid (DMSA) was modified onto the surface of the manganese oxide nanoparticles by the ligand exchange method (labeled as D-MnO). Finally, the D-MnO nanoparticles were cross-linked together by sulfhydrylated polyethylene glycol (HS-PEG-SH) through disulfide bonds to obtain cross-linked manganese oxide nanoassemblies (C-MnO). Transmission electron microscopy (TEM) analysis revealed that the MnO-OA nanoparticles exhibited approximately spherical morphology with excellent dispersibility, displaying an average diameter of approximately 10 nm. (Fig. 1b). High-resolution transmission electron microscopy (HR-TEM) images revealed well-defined crystallinity, with a measured lattice spacing of 0.251 nm, matching the expected spacing for the (111) plane of MnO-OA nanoparticles (Figure S1a). The selected area electron diffraction (SAED) patterns of MnO-OA nanoparticles showed polycrystalline rings (Figure S1b), and obvious characteristic peaks are also observed in the X-ray diffraction (XRD) pattern (Figure S1c). These results indicate that the prepared MnO-OA nanoparticles mainly have a monocrystalline structure, and identical peak positions of XRD pattern were also consistent with the standard pattern of MnO (PDF#75–0626), which further confirmed the crystalline nature of the MnO-OA nanoparticles. After surface modification with DMSA, the D-MnO nanoparticles exhibited excellent monodispersity, and the size observed under the TEM did not show any significant change (Fig. 1c). Meanwhile, the Fourier transform infrared (FT-IR) spectrum reveals the absence of characteristic peaks at 2928 cm⁻¹ and 2854 cm⁻¹, corresponding to the asymmetric and symmetric stretching vibrations of the methylene (-CH₂) groups in the unsaturated chain of oleic acid (OA). This confirms the complete removal of OA and the successful coating of DMSA (Figure S2). Under TEM observation (Fig. 1d), it was found that under the action of HS-PEG₂₀₀₀-SH, the D-MnO nanoparticles were cross-linked with each other through disulfide bonds to form aggregated assemblies (C-MnO). Through dynamic light scattering (DLS) tests, it was known that their average size increased from approximately 15 nm to about 342 nm (Fig. 1e). A scanning TEM with high-angle annular dark-field (HAADF-STEM) image further affirms that C-MnO was a closely arranged nanocluster (Fig. 1f). The corresponding elemental mapping (Fig. 1g) and a spectrum of energy dispersive X-ray spectroscopy (EDX) (Figure S3) showed the existence and distribution of all the expected elements (Mn, O, and S) of C-MnO. In order to further analyze the chemical composition and elemental state of C-MnO, X-ray photoelectron spectroscopy (XPS) analysis was carried out, the peaks at 640.7, 652.7, 530.4 and 162.9 eV are attributed to the binding energies of Mn 2p_3/2, Mn 2p_1/2, O 1s and S 2p (Figure S4). Meanwhile, the Raman spectrum of C-MnO shows a characteristic peak at 522.38 cm⁻¹ which is corresponding to the disulfide bond (Figure S5) [45]. The experimental results conclusively demonstrate the successful synthesis of C-MnO nanoassemblies, as evidenced by the comprehensive characterization data.

Fig. 1.

Synthesis, characterization, and responsive performance of C-MnO. (a) Schematic illustration of the synthetic pathway for C-MnO; TEM images of (b) Mn-OA, (c) D-MnO and (d) C-MnO; (e) Dynamic light scattering measurements of Mn-OA, D-MnO and C-MnO; (f) TEM image, (g) scanning transmission electron microscopy with high-angle annular dark-field (HAADF-STEM) images and corresponding element mapping of C-MnO; (h) Schematic illustration of the response mechanism between C-MnO and GSH; (i) TEM images of C-MnO after incubated with GSH, and the inset images display the hydration diameters; (j) MALDI-TOF mass spectra (m/z) of C-MnO after reacting with GSH; (K) T₁ of C-MnO treated with different concentrations of GSH (0–10 mM); (l) Pseudocolor T₁- weighted MR images and (m) corresponding T₁ relaxation rate analysis of C-MnO at 0.5 T. (n) The selective response of T₁ for C-MnO incubated with various species. n = 3

Subsequently, we assessed the in vitro responsiveness of C-MnO to GSH. Upon GSH addition, C-MnO underwent decrosslinking (Fig. 1h), resulting in a size reduction to approximately 1 nm (Fig. 1i). Concurrently, the solution transitioned from light yellow to colorless, indicating the degradation of the C-MnO nanoassemblies. Meanwhile, UV-Vis spectroscopy revealed the disappearance of the absorption peak (Figure S6), accompanied by a rapid increase in the T₁ MRI signal, which subsequently stabilized over time (Figure S7a-b). Collectively, these findings suggest that under some special physiological and pathological conditions—particularly in GSH-rich environments such as the liver (where GSH levels reach ~ 10 mM)—C-MnO is likely to undergo in vivo decrosslinking/degradation, thereby triggering T₁ MRI signal activation [30]. Subsequently, we further analyzed the degradation mechanism of C-MnO through matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry. The results identified the presence of [GSH + Mn²⁺ − H⁺]⁺ in the C-MnO solution containing GSH and observed its significant peak (Fig. 1j). These results indicate that during the degradation process, C-MnO first disassembles into monodispersed MnO nanoparticles, and further disintegrates and releases Mn²⁺. These released Mn²⁺ are rapidly chelated by the excess GSH in the solution to form Mn²⁺-GSH chelates. We also studied this degradation process by measuring the changes in T₁ relaxation time of C-MnO under the condition of adding different concentrations of GSH. As expected, the T₁ value of C-MnO gradually decreased with the increase of GSH concentration and then tended to stabilize (Fig. 1k). This is due to an excess of GSH, and the C-MnO in the solution has been completely degraded. Meanwhile, with the increase of GSH concentration, the gradual disintegration of C-MnO can be visually observed (Figure S8a-c). Since the presence of Mn²⁺-GSH chelates significantly enhanced the T₁ signal of MRI, we subsequently investigated the feasibility of C-MnO for redox-activated T₁-weighted MRI. Compared with the situation without GSH, the T₁-weighted MR images of C-MnO containing GSH show a strong T₁ MRI signal, which is enhanced with the increase of C-MnO or GSH concentration (Fig. 1l and Figure S9). The longitudinal relaxation rate of C-MnO containing 1 and 10 mM GSH was calculated to be 4.58 and 10.12 mM⁻¹s⁻¹ (Fig. 1m), which was several times higher than that of C-MnO without GSH. Meanwhile, under the clinical magnetic field of 3.0 T, the longitudinal relaxation rate before and after the C-MnO response with GSH also increased from 0.20 mM⁻¹s⁻¹ to 3.03 mM⁻¹s⁻¹ (with 1 mM GSH) or 4.78 mM⁻¹s⁻¹ (with 10 mM GSH) (Figure S10a-b). This indicates that C-MnO can be used as a GSH-responsive T₁ MRI nanoprobe. Regarding the stability of C-MnO, a series of evaluations have shown that C-MnO exhibits excellent stability under physiological conditions, the hydrated particle size and Zeta potential of C-MnO remained relatively stable for about one week in water or 10% FBS [46, 47]. (Figure S11a-c). Furthermore, an in vitro test was conducted to evaluate the selectivity of C-MnO for GSH. The results showed that the T₁ relaxation time was significantly shortened only in the presence of GSH (Fig. 1n), indicating its high specificity and anti-interference ability. This implies that the abundant GSH in normal liver can provide the optimal conditions for the activation of C-MnO’s T₁ MRI signal, and it can serve as an important tool for visualizing the GSH content in the liver.

Response performance of C-MnO at the cellular level

Due to the excellent performance of C-MnO in vitro, we further evaluated the behavior of C-MnO at the cellular level. Firstly, tests for cytotoxicity and cell uptake were conducted, and it was found that C-MnO had no significant effect on THLE-3 (normal liver cell) and 293T cells (normal kidney cell) below 10 µg/mL Mn (Figure S12a). The hemolysis assessment indicated that no significant hemolysis was observed even when the Mn concentration reached as high as 160 µg/mL (Figure S12b). The above experimental results indicate that C-MnO has excellent biocompatibility. In addition, the internalization of C-MnO in THLE-3 cells (normal liver cell) was monitored by confocal laser scanning microscopy (CLSM) and flow cytometry. The fluorescence intensity of C-MnO-treated cells increased significantly with prolonged incubation time, demonstrating a clear time-dependent internalization behavior (Figure S13a). Flow cytometry analysis confirmed similar internalization patterns to those observed by CLSM image (Figure S13b). The liver serves as the primary site for GSH synthesis, and L-BSO is a potent inhibitor of GSH synthesis [10, 48]. As shown in Fig. 2a, glutathione reductase (GR), a key enzyme in the GSH redox cycle, uniquely catalyzes the conversion of oxidized glutathione (GSSG) to reduced glutathione (GSH) [49]. Consequently, GSH depletion triggers a compensatory upregulation of GR to maintain cellular redox homeostasis. When GSH levels decrease, the degradation of C-MnO is inhibited, and this results in insufficient activation of the T₁ MRI signal, placing it in an “OFF” state. Conversely, increased GSH levels initiate C-MnO degradation, leading to efficient T₁ MRI signal activation and an “ON” state. To prove the above process, we pre-incubated THLE-3 cells with L-BSO. It was indeed observed that the intracellular GSH level decreased (Fig. 2b and Figure S14a-b), while the GR level increased with the increase of L-BSO concentration (Fig. 2h). Subsequently, to evaluate the MRI response of C-MnO in hepatic cells, THLE-3 cells pretreated with varying concentrations of L-BSO were incubated with C-MnO and subjected to cellular-level MRI assessment under a 0.5 T magnetic field. T₁-weighted MR images showed that the T₁ MRI signal gradually weakened with the increase of L-BSO concentration (Fig. 2c). This result was confirmed by quantitative analysis, and the signal-to-noise ratio (SNR) of T₁-weighted MR images in the control group was 2.4 times higher than that in the 600 µM L-BSO treatment group (Fig. 2d). Furthermore, to more clearly visualize the cellular dynamics of material disintegration in response to altered GSH levels, we co-incubated C-MnO with THLE-3 cells that have been pre-treated with either L-BSO or exogenous GSH. Our experimental results demonstrated that L-BSO treatment induced a significant decrease in intracellular GSH levels (Fig. 2e-f) accompanied by a concomitant increase in GR activity (Fig. 2i). Under these conditions, C-MnO maintained an aggregated state within THLE-3 cells (Fig. 2g). These findings strongly suggest that L-BSO-mediated GSH depletion prevents both C-MnO depolymerization and subsequent T₁-weighted MRI signal activation. Upon GSH restoration, we observed a corresponding decrease in GR expression along with intracellular disintegration and diffusion of C-MnO (Fig. 2g). The above results indicate that GSH is negatively correlated with GR levels, and the “activation” of the T₁ MRI signal of C-MnO depends on GSH.

Fig. 2.

Response performance of C-MnO at the cellular level. (a) Schematic illustration showing C-MnO as activated T₁ CE-MRI probes for the visual monitoring of GSH content; (b) The GSH contents in THLE-3 cells treated with L-BSO; (c) T₁-weighted MR images of THLE-3 cells treated as indicated; (d) SNR analysis corresponding to (c); Confocal laser scanning microscopy (CLSM) observation of (e) the changing process of GSH content and (g) disintegration of C-MnO in THLE-3 cells with same concentrations (10 µg/mL) of C-MnO under different conditions (Monochlorobimane, a kind of GSH fluorescent probe; Dil, a kind of fluorescent probe for cell membranes); (f) The mean fluorescence intensity of THLE-3 cells incubated with C-MnO corresponding to (e); (h) The GR contents in THLE-3 cells treated with L-BSO; (i) GR content corresponding to (e-g). All the data are reported as mean ± SD. n = 3. ^*P < 0.05; ^**P < 0.01; ^***P < 0.001

Ex vivo detection of liver injury with clinical methods

Extensive literature evidence indicates that prolonged or excessive use of INH or APAP can induce liver injury through the following mechanisms: primarily by triggering oxidative stress and inflammatory responses in hepatocytes, and secondarily by depleting hepatic GSH reserves. These pathological alterations collectively impair normal hepatocellular functions, ultimately leading to drug-induced liver injury [22, 50]. In current clinical practice, the diagnosis of liver injury is predominantly based on two approaches: invasive liver biopsy procedures and the assessment of serum biochemical markers. Hematoxylin and eosin (H&E) staining showed that the hepatocytes of mice in the control group had normal morphology, complete structure, clear and intact nucleoli, and no obvious degeneration, necrosis, inflammatory cell infiltration or hepatic sinusoidal congestion. In the acute liver injury model, mild edema occurred in hepatocytes, and some cells showed steatosis but in a small range. In the severe liver injury model, the range of hepatocyte steatosis expands, the degree intensifies, and obvious necrosis of hepatocytes occurs. However, after administration of GSH, the degree of liver injury in each group was significantly reduced, showing a protective effect of GSH on hepatocytes (Fig. 3a). These results demonstrate both the successful generation of graded drug-induced liver injury models and the hepatoprotective efficacy of GSH through oxidative stress mitigation and necrosis reduction. Based on the above results, considering that liver biopsy is an invasive examination, we also adopted the clinically non-invasive and efficient blood biochemical test to measure the key indicators of liver function: alanine aminotransferase (ALT)、aspartate aminotransferase (AST) and alkaline phosphatase (ALP). As shown in Fig. 3b-c, the liver function indicators of some liver injured mice were higher than those of normal mice, and this result was consistent with the pathological examination. However, the liver function indicators in some liver injured mice exhibited abnormalities that were inconsistent with the pathological examination. These discrepancies are highlighted by red square (Fig. 3b) and blue circular (Fig. 3c) icons. This suggests that liver function indicators lack sufficient sensitivity to liver injury and are susceptible to significant fluctuations due to individual variations (e.g., gender, age) or concurrent diseases, potentially leading to false-positive or false-negative results. Additionally, these indicators cannot reflect real-time changes in liver injury, highlighting their diagnostic limitations. Thus, more effective strategies are needed to enable real-time monitoring and accurate grading of liver injury.

Fig. 3.

Ex vivo detection of liver injury with clinical methods. (a) Histological analysis of liver tissues, mice were treated with different drugs as indicated (PBS as control), and sections were stained with hematoxylin and eosin, scale bar represents 50 μm; (b-c) The main liver function indicators (ALT, AST, and ALP) in the serum from mice treated as indicated. The red square icon represent that the results of the indicators are abnormal compared with the normal group. The blue circular icon represents abnormal results when comparing the indicators with the corresponding injury groups. (d) The content of GSH in the liver tissues of mice after treatment with different drugs. (e) The GR content in the liver tissues of mice after being treated with different drugs. All the data are reported as mean ± SD. n = 3. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001

Evaluation of GR levels in liver during acute and severe liver injury

Glutathione reductase (GR) is a crucial enzyme in the antioxidant system of the liver, and its main function is to reduce oxidized glutathione (GSSG) to its reduced form (GSH), thereby maintaining critical intracellular redox balance [51]. Upon liver injury, ROS are initially overproduced within hepatocytes, inducing oxidative stress [52]. GR will be rapidly activated to cope with oxidative stress, and the increase in its activity may occur earlier than obvious hepatocyte injury or cell membrane rupture. Traditional liver function indicators (such as ALT, AST and ALP) are enzymes within liver cells and are usually released in large quantities into the bloodstream only after severe damage to liver cells (such as cell membrane rupture) [14]. Consequently, GR demonstrates greater sensitivity than conventional liver function markers in detecting early liver injury, owing to its liver-specific activity profile and rapid oxidative stress response. This positions GR may as a superior biomarker for timely diagnosis of initial liver injury. In order to clarify the correlation between GSH and GR content during liver injury, we measured the contents of GSH and GR in the liver tissues of mice treated with different drugs. The results in Fig. 3d-e showed that GR in the liver tissues of normal mice maintained the basal activity level. However, with the aggravation of liver injury, the activity of GR in the liver increased significantly in a dose-dependent manner, while the content of GSH showed a negative correlation with the content of GR. This indicates that the dynamic change trends of GR and GSH during the process of liver injury are highly correlated with the degree of liver injury. The GSH regulatory pathway mediated by GR can fill the gap in the monitoring of early liver injury and the occurrence of “false positive” or “false negative” results. The GR activity in the liver significantly increases during the early oxidative stress stage when hepatocytes have not ruptured, earlier than the change window of liver function indicators [53–57]. Due to the significant differences in GR content changes at the cellular level among different treatment groups, in order to further verify the changes in GR content during liver injury and repair at the in vivo level, we used immunofluorescence technology to detect the distribution of GR during the injury and repair stages of liver. As shown in Fig. 4a-b, in liver injury caused by INH or APAP group, the fluorescence intensity of GR gradually increases with the aggravation of the injury degree. However, after supplementing GSH to each group, the fluorescence intensity of GR was significantly decreased compared with the corresponding injury group. The quantification of fluorescence intensity also further confirmed these observations (Fig. 4a-b). These findings suggest that cellular oxidative stress triggers GR overexpression as a compensatory mechanism to bolster cellular reducing capacity and maintain redox homeostasis. Consequently, GR expression levels directly correlate with hepatic antioxidant potential and provide a comprehensive reflection of liver functional status and injury progression. This positions GR as a sensitive biomarker for the early clinical detection of drug-induced liver injury, and it also means that GSH, which is directly related to GR, can also serve as an important biomarker for liver injury. The magnetic resonance probe based on the GR-GSH pathway we designed can specifically and indirectly monitor the GR activity in the liver by responding to the GSH level in the liver, thereby accurately capturing the biological signals of early liver injury.

Fig. 4.

Evaluation of GR levels in liver during acute and severe liver injury. Liver tissues were pretreated with (a) INH and (b) APAP, and treated with PBS as the control group for GR immunofluorescence staining imaging. (c-d) The mean fluorescence intensities of GR immunofluorescence corresponding to (a) and (b), respectively. Scale bar represents 20 μm. n = 3

In vivo CE-MRI detection of liver injury induced by INH or APAP

Prior to in vivo CE-MRI studies, we systematically evaluated the biosafety profile of C-MnO. No obvious lesions were found in the optical images of hematoxylin and eosin (H&E) stained tissue sections (heart, liver, spleen, lung and kidney) harvested 7 and 14 days after intravenous injection of C-MnO in Balb/c mice (Figure S15), demonstrating excellent biocompatibility of the C-MnO. Simultaneous in vivo fluorescence imaging revealed that C-MnO mainly accumulated in the liver, with the signal intensity reaching its peak 2 h after injection and then gradually decreasing (Figure S16a). This indicates that the metabolic process in the liver began 2 h after C-MnO injection, and fluorescence quantitative analysis further confirmed this distribution pattern (Figure S16b). Based on these findings, we established 2 h post-injection as the optimal imaging window for CE-MRI detection in vivo. To assess the potential of C-MnO as a nanoprobe for CE-MRI detection of drug-induced liver injury, we established murine models of liver injury using INH (150 mg/kg) or APAP (300 mg/kg) in Balb/c mice, with PBS-treated mice serving as controls. Acute and severe liver injury models were generated following standardized pathological protocols (Fig. 5a-b), mirroring our previous experimental conditions. The liver injury effects of INH and APAP are mediated through oxidative stress mechanisms. Under physiological conditions in liver, GSH serves as a primary cellular antioxidant by directly scavenging free radicals and ROS [58]. Due to the explosive increase of ROS in severe liver injury, a large amount of GSH is consumed to counteract oxidative stress, resulting in a sharp decrease in its content. Meanwhile, under conditions of severe liver injury, due to excessive accumulation of drugs in the liver, forces the liver to invoke more GSH for detoxification, thereby accelerating its consumption. In contrast, acute liver injury results in only modest GSH reduction while maintaining partial hepatic compensatory mechanisms. Therefore, after intravenous injection of C-MnO, the liver region of mice pretreated with PBS showed that the T₁ MRI signal gradually increased over time (Fig. 5c). It indicates that C-MnO responds to the abundant GSH in the liver, resulting in a significant enhancement of T₁ MRI signal. Through quantitative calculation, it can be determined that the ratio of the SNR (SNR_Post/Pre) of the liver in the control group 2 h after intravenous injection of C-MnO to the SNR of the liver before injection is 2.27. In the acute liver injury group, due to the mild decrease of intracellular GSH level, the increase of T₁ MRI signal in the INH or APAP acute liver injury groups slowed down (Fig. 5c), with SNR_Post/Pre ratios of 1.42 and 1.31, respectively (Fig. 5d). However, in cases of severe liver injury with extensive hepatocellular necrosis, combined oxidative stress and substrate limitation for GSH synthesis result in profound GSH depletion. Therefore, the signal change in the severe injury group was extremely weak, and the SNR_Post/Pre ratio dropped to 1.06 (Fig. 5d). After supplementing with GSH, the T₁ MRI signal increased in both the acute liver injury group and the severe liver injury group (Fig. 5c-d), which indicates that GSH alleviates liver injury caused by INH or APAP. By comparing the T₁ MRI signals of the normal liver and the graded injury liver and combining them with their pathological sections, it was found that the changes of T₁-MRI signals were closely related to the degree of liver injury, especially the values of SNR_Post/Pre can directly and intuitively reflect the degree of liver injury. The more severe the liver injury, the lower the T₁ MRI signal or the values of SNR_Post/Pre. Finally, after the scanning was completed, we conducted liver histopathological examinations (Figure S17) and serum biochemical tests (Figure S18a-b) on these mice. The results showed that there was no significant difference in H&E staining of liver tissue before and after probe administration, indicating that the probe had basically no effect on the process of liver injury and had low hepatotoxicity. In contrast, the blood biochemical indicators were in abnormal states both after and before the probe administration, making it difficult to objectively reflect the real-time dynamic changes of liver injury, indicating that this detection method has significant application limitations. The above results demonstrate that the C-MnO nanoprobes can be used as an important tool for real-time visual monitoring, assessment and grading diagnosis of drug-induced liver injury in CE-MRI. This will provide a new imaging-based supplementary strategy for the detection of drug-induced liver injury in clinical practice, and has great potential for clinical application.

Fig. 5.

In vivo CE-MRI detection of liver injury induced by INH or APAP. (a-b) Schematic illustration for injection and imaging procedure, and we defined the time of C-MnO injection as “0 min”; (c) T₁ CE-MRI of mice treated with different drugs as indicated (PBS as control); (d) SNR analysis corresponding to (c). n = 3

Conclusions

In conclusion, we successfully designed and constructed a GSH-activated T₁-type CE-MRI nanoprobe (C-MnO) for the graded diagnosis of liver injury in mice. Meanwhile, it was verified that GR can be used as a sensitive indicator for the auxiliary diagnosis of drug-induced liver injury. Owing to its high disulfide bond content, C-MnO undergoes GSH-mediated disassembly and degradation upon entering the normal mouse liver (where GSH is abundant), leading to the activation of the T₁ MRI signal. In the injured mouse liver, oxidative stress depletes GSH, causing the T₁ MRI signal to remain in the “OFF” state. This enables the C-MnO nanoprobe to accurately detect INH- or APAP-induced graded liver injury of mice within 2 h. It is worth noting that the T₁ MRI signal intensity of the liver in mice after intravenous injection of C-MnO was negatively correlated with the severity of liver injury. Furthermore, C-MnO allows real-time, non-invasive monitoring of dynamic T₁ MRI signal changes during liver injury in live mice via CE-MRI. When integrated with complementary histopathological and biochemical analyses, this multimodal approach offers three key advantages: (1) enhanced sensitivity for early injury detection, (2) comprehensive evaluation of injury etiology and severity grading, and (3) precise development of personalized therapeutic regimens. This strategy not only addresses current limitations in clinical diagnosis of liver injury but also demonstrates significant potential for clinical translation, inspiring further development of new visualization-based diagnostic approaches.

Experimental section

Synthesis of manganese oleate (Mn-OA)

Sodium oleate (24.36 g) was dissolved in a mixture of 30 mL of ethanol and 40 mL of distilled water at room temperature. Subsequently, a mixed solution of manganese chloride tetrahydrate in 10 mL of distilled water was added. The resulting mixture was heated to 70 °C and maintained at this temperature overnight. After 24 h, n-hexane was introduced, and the reaction was allowed to proceed for an additional 24 h. The solution was then transferred to a separatory funnel, and the upper organic layer containing the Mn-oleate complex was washed several times with distilled water.

Synthesis of manganese oxide nanoparticles (MnO-OA)

1-Hexadecene (13 mL) was degassed at 100 °C for 10 min under vacuum to remove residual water and oxygen. The solution was then heated to 280 °C under a nitrogen atmosphere, followed by the addition of a mixture containing 1-hexadecene (2 mL) and manganese oleate (0.62 g). The reaction was maintained at this temperature for 10 min to ensure sufficient nanoparticle growth. After the reaction, the mixture was cooled to room temperature, during which the solution color transitioned from light green to dark brown. The nanoparticles were then precipitated by adding acetone (50 mL), isolated via centrifugation at 12,000 rpm for 10 min, and finally redispersed in hexane, a non-polar solvent.

Surface modification of manganese oxide nanoparticles (D-MnO)

The hydrophobic manganese oxide nanoparticles were rendered hydrophilic via ligand exchange with meso-2,3-dimercaptosuccinic acid (DMSA). Briefly, the nanoparticles were dispersed in a mixture of toluene (10 mL) and n-hexane (2 mL), followed by the addition of a DMSA solution (0.1 g dissolved in 5 mL methanol). The reaction mixture was stirred at room temperature for 24 h, after which the nanoparticles were transferred to PBS. Finally, the hydrophilic manganese oxide nanoparticles (D-MnO) were isolated by filtration through a 220 nm membrane.

Crosslinking of manganese oxide nanoparticles (C-MnO)

The sulfhydryl groups of the hydrophilic manganese oxide nanoparticles (D-MnO) were activated using 2,2’-dithiodipyridine (DTDP). First, a saturated DTDP solution was prepared by dissolving DTDP (15 mL) in phosphate-buffered saline (PBS) at 60 °C under continuous stirring until no solid residue remained. The solution was then cooled to room temperature and filtered through a 220 nm membrane to obtain the saturated DTDP solution. The activated D-MnO nanoparticles were added to the saturated DTDP solution and stirred overnight at room temperature. After the reaction, the nanoparticles were washed to remove unreacted DTDP in the supernatant. Subsequently, the nanoparticles were dispersed in a solution of thiol-PEG-thiol (0.05 g) in PBS and stirred for 8 h at room temperature to facilitate cross-linking. Finally, the nanoparticles were redispersed in PBS to obtain the cross-linked manganese oxide nanoparticles (C-MnO).

Evaluation of GSH and GR content in THLE-3 cells

THLE-3 cells were seeded in petri dishes at a density of 5 × 10⁶ per dish and cultured for 12 h. For GSH content, the cells were treated with PBS or L-BSO (GSH inhibitor, 200 µM, 400 µM and 600 µM) for 12 h. Then the cells were harvested, washed two times with PBS, re-suspended in 1 mL reagent one and subjected to ultrasonic disruption in an ice bath. After centrifugation, the supernatants were collected and added the corresponding reagent according to the steps of the kit for testing. For GR content, THLE-3 cells were seeded in petri dishes at a density of 5 × 10⁶ per dish and cultured for 12 h. The subsequent processing steps were consistent with the detection of GSH content. For the pretreatment group, the cells were treated with PBS, L-BSO (200 µM and 600 µM) and GSH (1 mM and 10 mM) for 12 h. The cells were incubated with C-MnO (10 µg/mL) at 37 °C for 2 h. Then the cells were harvested, washed two times with PBS, re-suspended in 1 mL reagent one and subjected to ultrasonic disruption in an ice bath. After centrifugation, the supernatants were collected and added the corresponding reagent according to the steps of the kit for testing. For CLSM observations, the pretreated cells were stained with Dil (1 µM) for 20 min and stained with DAPI for 10 min at 37 °C in the dark, and washed three times with PBS. Finally, the cells were imaged using a confocal microscope.

Cellular MRI

THLE-3 cells were seeded in petri dishes at a density of 1 × 10⁷ per dish and cultured for 24 h. The cells were treated with PBS or L-BSO (200 µM, 400 µM and 600 µM) for 12 h. The cells were incubated with C-MnO (10 µg/mL) at 37 °C for 6 h. Then the cells were harvested, washed three times with PBS, and transferred into 200 µL centrifugal tubes. MR images were acquired on the aforementioned 0.5 T MR scanner with the following parameters: Spin Echo (SE) sequence, Repetition Time (TR)/Echo Time (TE) = 200/16.14 ms, Number of scans (NS) = 8, Field of View (FOV) = 25 × 25 mm. The signal to noise ratios (SNR) were evaluated by soft “ImageJ”.

In vivo MRI experiment

Prior to MRI, mice were anesthetized by intraperitoneal injection of 4% chloral hydrate. In acute liver injury models, severe liver injury models and normal mice, T₁-weighted images (T₁-WI) were acquired on the aforementioned 3.0 T MR scanner before and after intravenous injection of C-MnO at a dosage of 5 mg/kg body weight. The parameters were as follows: TR/TE = 500/13 ms, slice thickness = 1 mm, slice spacing = 0 mm, FOV = 40 × 40 mm, matrix = 256 × 256.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

K.L. and C.W. conceived the idea. K.L., C.W. and N.Y. designed the experiments. N.Y., C.W. and J.W. performed the experiments. K.L., Z.C. and H.L. analyzed the data. Y.W. and C.B. contributed to cell culture. C.S. helped with MALDI-TOF mass spectra analysis. Z.C., J.X., Q.C., H.C. and S.G. helped with the MRI experiments. N.Y., K.L., C.W., Z.C. and H.L. co-wrote the paper. All authors discussed the results and commented on the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (82302179, 22477104 and 92578118), the Youth Top-notch Talent Program of the Fujian Young Eagle Plan, the Youth Innovation Team Plan of Colleges and Universities in Shandong Province (2023KJ368), the Health and Wellness Innovation Team Construction Project of Shandong Province, the Natural Science Foundation of Shandong Province (ZR2024QB409 and ZR2024QH251), the Natural Science Foundation of Fujian Province (2023J06002), and the Binzhou Medical University Research Initiation Fund (50012304640).

Data availability

The datasets during and/or analysed during the current study available from the corresponding author on reasonable request.The datasets supporting the conclusions of this article are included within the article and its additional files.

Declarations

Ethics approval and consent to participate

Animal Ethics: all the animal experiments in our study were in accordance with the Ethics Committee Guidelines of Binzhou Medical University.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.