A magnetic nanorobot with guided aggregation and thrombin-triggered coagulation for endovascular embolization

Abstract

Current endovascular embolic agents face challenges in achieving precise localization and targeted embolization. Herein, we report a novel magnetic nano-embolization robot (MNER) that integrates real-time magnetic aggregation with thrombin-triggered coagulation, enabling a transition from passive diffusion to magnetically controlled focal embolization. The MNER features a core-shell structure with a magnetic core and a thrombin outer layer. Upon endovascular injection, MNER aggregates at the target sites under magnetic guidance, while surface-anchored thrombin triggers fibrin polymerization, promoting the rapid formation of stable blood clotting and localized embolization. The MNER exhibits excellent magnetic response in static and circulating fluid environments in vitro. In rabbit models, the MNER achieved effective and sustained embolization in renal arteries (≈3 mm) without migration or recanalization, and enabled precise occlusion in hemorrhagic mesenteric arteries with complex branches. Owing to its easy usage, biocompatibility, and potent embolization, the MNER represents a promising next-generation platform for endovascular embolization, enabling to achieve precise local retention at the target site via magnetic aggregation in conjunction with catheter-based delivery.

Graphical abstract

Highlights

• •Core-shell MNER enables local aggregation under magnetic field for precise embolization.
• •Magnetic aggregation and thrombin-triggered coagulation for rapid, localized embolization.
• •Complete, sustained occlusion in renal and complex mesenteric arteries without recanalization.
• •Excellent biocompatibility, easy handling, and potent efficacy for next-generation embolization.

1. Introduction

Transcatheter arterial embolization is a critical treatment in interventional medicine, and its therapeutic effect mainly depends on the performance of embolic agent [1,2]. Clinical embolic agents can be classified into two categories: liquid agents [such as n-Butyl Cyanoacrylate (NBCA), Onyx (composed of ethylene-vinyl alcohol copolymer, tantalum particles and dimethyl sulfoxide), and precipitating hydrophobic injectable liquid (PHIL)] and solid agents (metal coils and particulates) [3,4]. Liquid agents rely on solvent exchange or in situ polymerization to undergo phase transformation, thereby achieving embolization [5]. However, NBCA undergoes exothermic polymerization reaction (80 – 90 °C) upon contact with blood, forming a glue in situ, which may damage blood cells, vessel walls, and surrounding tissues. Both Onyx and PHIL use dimethyl sulfoxide (DMSO) as a solvent, which can cause vasospasms as well as neurological and cardiovascular toxicity [3,6]. Solid agents depend on physical aggregation to achieve embolization and therefore can use water as a solvent. However, traditional solid agents relying solely on physical aggregation are prone to migration, leading to ectopic embolization [2,7].

In recent years, to address ectopic embolization of the traditional solid embolic agents, various particulate agents capable of forming stable aggregates in situ have been reported [8,9]. For instance, Ye and co-workers prepared amphiphilic multi-block polyurethane microspheres, which can fuse into a block plug in blood and achieve embolization [10]. Furthermore, pH-responsive microspheres have been engineered to encapsulate a small-molecule vascular disrupting agent along with thrombin for combined delivery [8]. This strategy promotes synergistic effects that trigger coagulation within blood vessels, thereby improving the therapeutic outcome of tumor-starvation approaches. However, these solid agents still rely on passive delivery mechanisms. Their distribution remains heavily influenced by hemodynamics and lacks targeting precision, making it difficult to achieve truly targeted embolization.

Recently, microrobots that can utilize magnetic fields, acoustic fields, or biochemical stimuli to achieve active navigation in biological environments have been rapidly developed [[11], [12], [13], [14], [15]]. Among them, magnetically controlled microrobots have attracted significant attention due to their high controllability and excellent biocompatibility [16,17]. Go et al. reported microrobots that can be visualized in real time using X-ray and magnetic resonance imaging (MRI) and magnetically guided to tumor-feeding vessels for transcatheter hepatic chemoembolization [12]. Xu and co-workers reported a magnetically driven biohybrid blood hydrogel fiber for personalized intracranial tumour therapy [18]. Zhang and co-workers report thrombolytic drug-anchored nanorobots that can accurately navigate to submillimeter-scale thrombotic sites and perform magnetically guided thrombolysis [19]. Zhao and co-workers use 3D printing to fabricate helical magnetic microrobots capable of traveling in pulsatile blood flow and treating thrombosis and aneurysms via combined negative pressure compression and drug delivery [20]. Meanwhile, surface functionalization and stabilization strategies for magnetic nanoparticles have become increasingly sophisticated, establishing a solid technological foundation for their biomedical applications [21]. Collectively, these systems integrate the precision of magnetic navigation with localized therapeutic actions, offering valuable inspiration for the design of next-generation embolic agents.

Herein, we develop a novel magnetic nano-embolization robot (MNER) with magnetic aggregation and thrombin-triggered coagulation for controllable and localized endovascular embolization. The MNER with a core-shell structure was constructed through a sequential four-step synthesis process, comprising silica encapsulation of a magnetic iron (II, III) oxide (Fe₃O₄) core, surface amination, bis (2,5-dioxopyrrolidin-1-yl) 4,7,10,13,16,19-hexaoxadocosanedioate (bis-PEG₆-NHS ester) crosslinker conjugation, and covalent immobilization of thrombin (Scheme 1A). Upon injection into blood vessels, the MNER first aggregates at the target site under magnetic guidance, exhibiting the core robotic capabilities of controllable motion and precise execution. Unlike conventional embolic agents, which passively distribute with the blood flow, MNER achieves spatially controllable distribution under an external magnetic field. Concurrently, surface-bound thrombin activates fibrinogen, leading to the formation of water-insoluble fibrin polymers that interweave into a network, entrapping red blood cells and resulting in localized blood clotting formation and embolization (Scheme 1C). In rabbit models, under the guidance of magnetic field, the MNER demonstrates effective and safe embolization in renal arteries (≈3 mm) without migration or recanalization. Furthermore, it achieves precise embolization in a hemorrhagic mesenteric artery model with complex branches. Benefiting from its easy usage, good biocompatibility, and effective embolization, the MNER represents a promising next-generation embolic agent for local retention via magnetic aggregation in conjunction with catheter-based delivery.

Scheme 1.

Schematic diagram of magnetic nano-embolization robot (MNER) with magnetic aggregation and thrombin-triggered coagulation for targeted endovascular embolization. (A) Structure of the MNER. (B) Transarterial embolization operation. (C) Targeted vascular embolization of the MNER.

2. Results and discussion

2.1. Fabrication and characterization of the magnetic nano-embolization robot (MNER)

The core-shell structured magnetic nano-embolization robot (MNER) was fabricated through a four-step hierarchical assembly strategy (Fig. 1A). First, the magnetic iron (II, III) oxide (Fe₃O₄) core was encapsulated with the tetraethyl orthosilicate (SiO₂) through a biphasic stratification method [19]. Subsequently, surface amination of Fe₃O₄@SiO₂ was performed using 3-aminopropyltriethoxysilane (APTES) to introduce reactive amine groups. Afterwards, bis (2,5-dioxopyrrolidin-1-yl) 4,7,10,13,16,19-hexaoxadocosanedioate ester (bis-PEG₆-NHS ester) underwent amidation reaction with Fe₃O₄@SiO₂-NH₂ to anchor succinimide ester (-NHS) groups on their surface. Finally, thrombin was covalently immobilized through the reaction between its amine (-NH₂) groups and the surface NHS esters on Fe₃O₄@SiO₂-NHS, and the final product designated as the magnetic nano-embolization robot (MNER).

Fig. 1.

Fabrication of the magnetic nano-embolization robot (MNER). (A) Schematic of the MNER preparation. (B) XRD of the Fe₃O₄ and Fe₃O₄@SiO₂-NH₂. (C, D) FTIR spectra of the Fe₃O₄, Fe₃O₄@SiO₂, Fe₃O₄@SiO₂-NH₂, Bis-PEG₆-NHS and Fe₃O₄@SiO₂-NHS. (E) Microscopic images of the fluorescent thrombin/Fe₃O₄ and fluorescent thrombin/Fe₃O₄@SiO₂-NHS. (F) Diameter distribution of the MNER. (G) TEM image and Sulfur element distribution of the MNER. (H) Zeta potential measurements of Fe₃O₄ and Fe₃O₄@SiO₂-NH₂ and MNER (n = 3). Data are shown as mean ± SD. Statistical significance was calculated by one-way ANOVA. ns, not significant, *p < 0.05, ****p < 0.0001.

Next, the chemical composition of the Fe₃O₄@SiO₂, Fe₃O₄@SiO₂-NH₂, and MNER were studied. X-Ray diffraction (XRD) was used to verify the synthesis of Fe₃O₄@SiO₂ and Fe₃O₄@SiO₂-NH₂. As shown in Fig. 1B, the pure Fe₃O₄ exhibited sharp crystalline peaks. The appearance of amorphous peaks in Fe₃O₄@SiO₂ and Fe₃O₄@SiO₂-NH₂ samples demonstrated the successful coating of SiO₂ on the surface of Fe₃O₄ nanoparticles [22]. Moreover, Fourier transform infrared spectroscopy (FTIR) was conducted. Compared to Fe-O peak (604 cm⁻¹) of the Fe₃O₄, Fe₃O₄@SiO₂ showed -Si-O (459 cm⁻¹ and 791 cm⁻¹) and O-Si-O (1086 cm⁻¹) peaks, confirming the formation of the Fe₃O₄@SiO₂ (Fig. 1C). The Fe₃O₄@SiO₂-NH₂ showed -C-N (1321 cm⁻¹) and -NH₂ (1562 cm⁻¹) peaks, confirming the grafting of the NH₂ groups. Compared to Fe₃O₄@SiO₂-NH₂, the new amide bond (1556 cm⁻¹ and 1645 cm⁻¹) and -C-H (2879 cm⁻¹ and 2931 cm⁻¹) peaks occurred in Fe₃O₄@SiO₂-NHS, confirming the grafting of the NHS groups (Fig. 1D). Besides, the fluorescent thrombin was mixed with Fe₃O₄@SiO₂-NHS and Fe₃O₄, respectively. After being washed with deionized water, samples were observed under a fluorescence microscope. The images revealed pronounced fluorescence in the MNER samples, whereas the Fe₃O₄ group exhibited almost no fluorescence. This result indicates the successful surface conjugation of thrombin to the nanoparticles (Fig. 1E). Moreover, to quantitatively characterize thrombin loading on MNER, the amount of immobilized thrombin was determined from the difference between the total amount of thrombin added during preparation and the amount of residual thrombin in the final solution. The results showed that the activity of immobilized thrombin was 6.2 ± 0.6 U/mg of MNER (n = 3). And the enzymatic activity of the MNER was 86.2 ± 1.4%, which was tested with a chromogenic substrate assay and compared with free thrombin at an equivalent enzyme amount (n = 3). Furthermore, to evaluate thrombin release stability during injection, we measured the amount of free thrombin in the solution. No free thrombin was detected in the MNER suspension either before or after injection, indicating that thrombin was not released from MNER during use.

In addition, dynamic light scattering (DLS) test revealed a hydrodynamic diameter of 371.6 ± 24.7 nm of the MNER, indicating good monodispersity (Fig. 1F). Meanwhile, transmission electron microscopy (TEM) images clearly reveal its core–shell structure (Fig. 1G). In a large field of view, multiple uniformly sized and morphologically consistent nanostructures can be observed (Fig. S1). Furthermore, the elemental analysis image showed that there was uniform sulfur (S) on the surface of MNER, which further indicated the successful grafting of thrombin. Finally, the zeta potential of the nanoparticles shifted from +12.7 ± 0.2 mV (Fe₃O₄) and +13.9 ± 0.4 mV (Fe₃O₄@SiO₂-NH₂) to +8.4 ± 0.5 mV for the MNER, a change primarily attributed to the charge shielding effect of the grafted thrombin (Fig. 1H).

2.2. Properties of the MNER for the endovascular embolization

Upon contacting blood, the thrombin on the surface of the MNER can actuate fibrin monomers to connect with each other to form water-insoluble fibrin polymers, which interweaved with each other into a network to trap red blood cells, resulting in forming blood clotting (Fig. 2A). To evaluate coagulation ability of the MNER, the normal saline, Fe₃O₄@SiO₂-NH₂, thrombin and MNER were respectively added to the anticoagulant blood. 2 min later, the thrombin and MNER groups formed blood clotting, while the blood in normal saline and Fe₃O₄@SiO₂-NH₂ groups was flowing (Fig. 2B). Moreover, when MNER was deposited into molds containing anticoagulated blood, stable blood clotting retaining the specific shapes of the molds were formed. This demonstrates its ability to conform to complex anatomical contours at a target site (Fig. 2C). Besides, in the concentration range of 10-20 mg/mL, the blood clotting time shortened with the increased MNER content. Subsequently, even if the concentration was further increased, the blood clotting time remained stable (43.7 ± 1.5 s) (Fig. 2D). Moreover, the storage modulus (G′) of the blood increased from 8.4 ± 0.1 Pa to 150.9 ± 4.6 Pa upon contacting MNER, indicating the formation of the stable blood clotting (Fig. 2E). Therefore, 20 mg/mL of MNER was chosen for the follow-up experiment.

Fig. 2.

Properties of the MNER for the endovascular embolization. (A) Schematic of coagulation mechanism of the MNER (left), and the SEM images of the aggregated red blood cells upon contacting MNER suspension (right). (B) Coagulation test of normal saline, Fe₃O₄@SiO₂-NH₂, thrombin and MNER. (C) Preparation of the blood clotting with different shapes. (D) Blood clotting time of the MNER with different concentrations (n = 3). (E) Time oscillatory sweeps of the anticoagulant blood (black) and blood clotting (red). (F) X-ray fluoroscopy and CT scanning images (left) and CT values (right) of the MNER containing varying concentration of iohexol (n = 3). (G) Injection force of MNER suspension (n = 4). Data are shown as the mean ± SD. Statistical significance was analyzed with one-way ANOVA. ns, not significant, *p < 0.05, ****p < 0.0001.

Moreover, to enable X-ray visualization, iohexol was physically mixed with MNER suspension. The computed tomography (CT) attenuation value increased proportionally with the iohexol concentration. At 35 wt% iohexol, the radiopacity of the MNER suspension was statistically equivalent to the commercial contrast agent (Omnipaque), confirming its suitability for intraoperative tracking under X-ray guidance (Fig. 2F). Injectability through clinically relevant was assessed by measuring delivery forces for MNER suspension in narrow-lumen catheters. The injection force of the MNER suspension was 4.58 ± 0.06 N (1.7-F), 0.97 ± 0.05 N (2.7-F), and 0.42 ± 0.02 N (4.0-F), respectively, all well below the 50 N clinical safety threshold reported in the literature (Fig. 2G and S2) [23]. Moreover, no clogging was observed during injection. These results demonstrate that MNER can be smoothly delivered through clinically standard microcatheters (1.7 to 4.0 F).

2.3. Magnetically responsive ability of the MNER

When a magnet was brought close to a uniform MNER suspension, the MNER rapidly moved toward the magnet and accumulated (Fig. 3A). Meanwhile, the MNER suspension was measured by vibrating sample magnetometer (VSM), which saturation magnetization was 3.2 emu/g, showing good magnetic response ability (Fig. 3B). Furthermore, the fluorescence-labeled MNER rapidly formed chain-like alignments under a magnetic field, further demonstrating good magnetically responsive ability (Fig. 3C). In addition, the motion of MNER in a static liquid environment under magnetic guidance was investigated, as it represents one of the most critical performance aspects for magnetically responsive robots in practical applications. The magnetic field strength was 15.8 mT, with a field gradient of 0.8 T/m, and the magnet was positioned 3 cm from the sample. Under these conditions, MNER exhibited a migration velocity of approximately 8 mm/s and a response time of less than 1 s (Fig. S3). And the MNER can follow an “SYSU” path under magnetic actuation (Fig. 3D and movie S1 to S4). Furthermore, even within a maze, the MNER was capable of reaching the target quickly and accurately via magnetic steering (Fig. 3D and movie S5).

Fig. 3.

Magnetically responsive ability of the MNER. (A) Distribution of MNER suspension without and with a magnet. (B) Magnetic hysteresis loop of the MNER. (C) Distribution of fluorescence-labeled MNER in solution without and with a magnetic field. (D) Magnetically controlled movement of the MNER in static liquid environment. (E) Schematic diagram of the magnetic aggregation experimental setup, including a 3D-printed model of Y-shaped blood vessel and aneurysm, along with fluid control and magnetic control systems. (F, G) Finite element simulations of MNER at a flow velocity of 0.2 m/s in a vascular branch model (F) and an aneurysm model (G). (H, I) Magnetically controlled movement of the MNER in 3D branched blood vessel (H) and aneurysm model (I) with a circulating fluid environment.

Next, we evaluated the magnetically controlled movement of MNER in a circulating fluid environment (flow velocity was 0.2 m/s and wall shear stress approximately 1.4 Pa) using in vitro three-dimensional (3D) models of a Y-shaped blood vessel (diameter: 4 mm), an aneurysm (diameter: 6 mm) and a branched blood vessel (diameter: 2 mm) (Fig. 3E). First, finite element simulation was employed to analyze the influence of the magnetic field on the movement of MNER in the branched blood vessel and aneurysm models. As shown in Fig. 3F, in the absence of a magnetic field, MNER followed the bloodstream into different vascular branches. When magnetic field was applied, MNER moved to the target vessel. In the aneurysm model, although the fluid velocity decreased significantly upon entering the aneurysm sac (with the maximum velocity observed at the neck), most MNER continued to move along the flow path in the absence of magnetic field. After applying a magnetic field, MNER successfully accumulated inside the aneurysm (Fig. 3G). Subsequently, to demonstrate the magnetically controlled motion in the 3D model, a 4-F catheter was placed at the vascular bifurcation. Under magnetic actuation, MNER accurately migrated to the target vessel against the flowing fluid (Fig. 3H and movie S6). Similarly, when the catheter was positioned at the aneurysm neck, MNER gathered inside the aneurysm under magnetic guidance (Fig. 3I and movie S7). In addition, to quantitatively evaluate the targeting efficiency, we collected the effluent from the target branch of the in vitro vascular model and quantified the MNER concentration. The results showed that the targeting efficiency of MNER was approximately 90.6 ± 1.0% (Fig. S4). In summary, these in vitro experiments demonstrated that MNER could be precisely accumulated within the target vessel under magnetic control.

On the other hand, to explore whether MNER-induced thrombi could be pharmacologically reversed, we conducted an in vitro thrombolysis assay using urokinase, a clinically approved thrombolytic agent. MNER-induced clots were incubated with either urokinase or normal saline at 37 °C for 40 min. As shown in Fig. S5, the urokinase-treated clots exhibited clear signs of fragmentation and dissolution, whereas no appreciable changes were observed in the saline-treated group. These findings suggested that MNER-induced thrombi remain responsive to urokinase-mediated thrombolysis and may therefore be amenable to pharmacological reversal when needed.

2.4. In vitro and in vivo biocompatibility of the MNER

To assess the cytocompatibility of MNER, a co-culture model was established using human umbilical vein endothelial cells (HUVECs). After 24 h of incubation, the cells in MNER and blank groups demonstrated spindle-shaped morphology without any abnormal morphology. Meanwhile, quantitative cell counting kit-8 (CCK-8) analysis confirmed no statistically significant difference (p > 0.05) between MNER and blank groups (Fig. 4A). Moreover, to study the hemocompatibility of the MNER, they were co-incubated with erythrocyte suspensions for 24 h at 37 °C (Fig. 4B). The hemolysis ratio of the MNER was 1.9 ± 0.9%, showing its acceptable hemolysis rate as embolic agents [24].

Fig. 4.

Biocompatibility of the MNER. (A) Live/dead staining images (left) and cell viability (right) of the HUVEC after co-culturing with the blank media or MNER for 24 h (n = 3). (B) Hemolysis images (left) and hemolysis ratio (right) of normal saline (NS), MNER, and deionized water (DIW) (37 °C, incubation time = 24 h, n = 3). (C) Schematic of assessing in vivo biocompatibility of MNER. (D) Change of the rat's weights (n = 3). (E) Blood routine (left) and serum biochemical (right) tests of rats before and after subcutaneous implantation of MNER (n = 3). RBC: red blood cell; WBC: white blood cell; PLT: platelets; ALB: albumin; TP: total protein; CREA: creatinine; ALT: alanine aminotransferase; AST: aspartate aminotransferase. (F) Thickness of the granulation tissue (left) and fibrous tissue (right) in the skin surrounding the MNER (n = 3). (G, H) Hematoxylin and eosin (H&E) staining (G) and Masson's staining (H) staining (original magnification ×5, × 10, × 40) images of the control tissue and tissue adjacent to MNER. Data are shown as the mean ± SD. Statistical significance was analyzed with Student's t-test or one-way ANOVA. ns, not significant, *p < 0.05, **p < 0.01, ***p < 0.001.

Next, the in vivo biocompatibility of the MNER was assessed by directly injecting their suspension into the subcutaneous tissue of rats (three animals at each time point) [25] (Fig. 4C). At pre-set time, we recorded the rats’ weight, and collected their blood samples and tissues around the MNER (n = 3). The weight of the experimental rats kept increasing steadily (Fig. 4D). Moreover, the blood routine (red blood cells, white blood cells and platelets) and serum biochemical test [albumin (ALB), total protein (TP), urea, creatinine (CREA), alanine aminotransferase (ALT), aspartate aminotransferase (AST)] results showed no statistically significant difference between experimental and blank rats (Fig. 4E).

Subsequent histological analysis of the tissues surrounding MNER was performed using hematoxylin and eosin (H&E) and Masson's staining. The H&E staining images showed a time-dependent resolution of inflammation, characterized by a decreasing presence of neutrophils (green arrow) and macrophages (black arrow), alongside a thinning of the granulation tissue over time (Fig. 4F and G). Masson's staining indicated that fibrous tissue (double-headed red arrow) hyperplasia was most pronounced on day 7 and progressively diminished thereafter, indicating a progressively resolving inflammatory response (Fig. 4F and H) [25,26]. Furthermore, histological examination revealed no signs of necrosis in the tissues adjacent to MNER. Consistent with this, H&E staining profiles of other major organs (such as the heart, brain, liver, lung, spleen, testis, and kidney) exhibited no notable discrepancies between the treatment and control groups, indicating that MNER did not trigger a systemic inflammatory response syndrome (Fig. S6). These findings collectively demonstrate good in vivo biocompatibility of the MNER, supporting its potential application in vascular embolization.

2.5. Renal artery embolic performances of the MNER

To evaluate embolic efficiency and safety, we used MNER to embolize renal artery of the New Zealand rabbit. First, a 4.0-F microcatheter was inserted into the right renal artery via femoral access under digital subtraction angiography (DSA) guidance. Subsequently, 0.5 mL iohexol was injected to confirm vascular patency (white arrow), establishing a baseline for subsequent embolization. Then, under the control of magnetic field, 0.1 – 0.2 mL of MNER suspension was deployed into renal artery with the diameter of 3.02 mm (Fig. 5A–C, and movie S8). Due to the control of magnetic field, the MNER completely aggregated in the renal artery. Upon contact with blood, the surface-bound thrombin triggered the coagulation cascade, leading to stable blood clotting formation within the renal artery. Post-embolization angiography confirmed successful occlusion, as evidenced by the absence of contrast opacification in the embolized renal artery and its branches (Fig. 5C).

Fig. 5.

Kidney embolic performance of the MNER. (A) Schematic of the renal artery embolization. (B) Schematic of the experimental design for evaluating the efficacy and safety of MNER used for embolization in rabbits. (C) DSA images of the kidney (white dotted circle) before and post embolization, X-ray fluoroscopy image of the MNER (blue arrow) in the embolized kidney. (D) NECT, CTA, CDFI, SMI, CEUS images of the embolized kidney before and after embolization at 7, 14 and 28 days. Peach or yellow dotted outline represents the kidney embolized with MNER. (E) CT scanning images of the heart, liver, lung and brain in the experimental rabbit. Rabbits, n = 3.

At predetermined intervals (7, 14, and 28 days) following the embolization procedure, the experimental rabbits underwent a series of follow-up imaging assessments (three animals at each time point). These included contrast-enhanced computed tomography (CECT), computed tomography angiography (CTA), color Doppler flow imaging (CDFI), superb microvascular imaging (SMI), and contrast-enhanced ultrasound (CEUS), as detailed in Fig. 6D. Both non-enhanced CT (NECT) and CTA demonstrated the absence of contrast inflow into the embolized kidney (peach dotted outline). Quantitative analysis further revealed that the CT enhancement values of the embolized kidneys remained below 10 HU at all postoperative time points, indicating sustained absence of blood perfusion and sustained vascular occlusion (Fig. S7). Furthermore, CDFI, SMI, and CEUS confirmed the lack of blood flow signals (yellow dotted outline) and revealed progressive volume reduction of the embolized kidneys (Fig. 5D and movie S9). These findings validated effective and persistent embolization (9/9; three animals at each time point) without recanalization (0/9; three animals at each time point).In addition, the CECT images of the whole body showed no abnormal densities or structural alterations in critical organs (brain, heart, liver, lungs) at day 28, which confirmed that the MNER remained in the target vessel without ectopic embolization (Fig. 5E).

Fig. 6.

Histopathological evaluation of normal and embolized renal arteries. (A) Gross morphological images of bilateral kidneys before and after embolization at predetermined time points. (B) H&E staining images of the renal parenchyma of normal and embolized kidneys (original magnification ×10, × 40). Blue arrow and green dotted outline indicate the renal tubules and glomeruli, respectively. (C) H&E staining images of normal and embolized renal arteries (original magnification ×10, × 40). Blue arrow, yellow double arrow, green arrow, and black double arrow indicate the arterial intima, arterial media, arterial externa, and whole layer of artery, respectively. Red triangle and green triangle indicate thrombus and organized thrombus, respectively. (D) Masson's staining images of normal and embolized renal arteries (original magnification ×10, × 40). Green arrow, yellow double arrow, pink arrow, and black double arrow indicate the arterial intima, arterial media, arterial externa, and whole layer of artery, respectively. Red triangle and green triangle indicate thrombus and organized thrombus, respectively. (E) Weight of the rabbit's embolized kidney at 0 day, 7 days, 14 days, and 28 days (n = 3). (F) Thickness of the renal arterial wall before and post embolization (n = 3). (G) Thickness of collagen fiber before and post embolization (n = 3). Data are shown as the mean ± SD. Statistical significance was analyzed with one-way ANOVA. ns, not significant, *p < 0.05, ***p < 0.001, ****p < 0.0001.

Next, the bilateral kidneys and the MNER-embolized renal artery were collected. The embolized kidney showed significant atrophy at 7, 14, and 28 days (Fig. 6A and E). Afterwards, H&E staining was performed on renal parenchyma before and after embolization (Fig. 6B). At 7 and 14 days post-embolization, although the glomeruli (outlined by green dashed lines) and tubules (blue arrows) remained their basic architecture, a significant reduction in nuclear density was observed, indicative of ischemia. Subsequently, these structures underwent progressive atrophy, showing extensive coagulative necrosis by day 28. Besides, the cross sections of the renal artery were assessed through H&E staining and Masson's staining (Fig. 6C and D). The non-embolized renal artery exhibited a clear lumen, arterial intima (blue arrow), arterial media (yellow double arrow), and arterial externa (dark green arrow). At 7 days post embolization, the lumen of the renal artery was filled with thrombus (red triangle). Afterwards, thrombus underwent organization to form organized thrombus (green triangle), which can be characterized by the ingrowth of granulation tissue and deposition of collagen fibers. This organized thrombus can ensure permanent vascular occlusion, effectively eliminating the risk of recanalization. Consequently, the ischemic condition induced a time-dependent, progressive reduction in the total thickness of the arterial wall (Fig. 6F). Moreover, the thickness of the arterial media gradually decreased between days 0 and 14, which may be associated with an overall thinning of the arterial wall. However, between days 14 and 28, the thickness of the arterial media increased, as demonstrated by Marson's staining, due to collagen fiber proliferation within the artery (Fig. 6G). Collectively, these results demonstrate that MNER achieved stable, permanent vascular occlusion without recanalization.

Finally, the vascular and systemic safety of the MNER in rabbit's renal artery was evaluated. The weight of the experimental rabbits kept increasing steadily (Fig. S8A). The blood routine (red blood cells, white blood cells and platelets) and serum biochemical test (ALB, TP, urea, CREA, ALT, AST) results showed no statistically significant difference between experimental and blank rabbits (Fig. S8B). Histological examination of the renal arteries revealed intact vascular wall architecture without evidence of necrosis or excessive inflammation, supporting the favorable biocompatibility of MNER (Fig. 6C and E). Whole-body contrast-enhanced CT imaging showed no abnormal densities or structural alterations in major organs, including the brain, heart, liver, and lungs, at day 28 (Fig. 5E). Consistently, H&E staining and histopathological scoring of major organs revealed no ischemic lesions or infarction (Fig. S9). Careful examination of the microvasculature in H&E-stained sections of the heart, liver, lungs, and testes further showed no evidence of fibrin thrombi, platelet aggregates, or particle emboli (Fig. S9). Moreover, histological analysis of the renal parenchyma indicated that MNER was confined to arterioles measuring approximately 100–1000 μm and did not migrate into smaller downstream vessels, thereby excluding distal migration within the target organ (Fig. S10). Finally, to further assess the risk of systemic coagulation, MNER was injected into the rabbit ear artery, and blood samples were collected at pre-set time (15 min, 1 h and 24 h) for coagulation analysis. Compared with the baseline values obtained before embolization, no significant changes were observed in any of the coagulation parameters (p > 0.05). These findings indicated that the thrombin activity of MNER was highly localized within the target vessel and did not trigger systemic coagulation (Fig. S11). Overall, MNER can effectively, permanently, and safely embolize the renal artery (3.02 mm) of the rabbits.

2.6. Magnetically controlled targeted embolization of the hemorrhagic small blood vessel with branches

To further demonstrate the magnetically controlled targeted embolization using the MNER, a model of hemorrhagic small blood vessels with branches was employed (Fig. 7A). A 2.2-F microcatheter was inserted into the mesenteric artery (MA) via carotid access under DSA guidance, followed by iohexol injection confirming vascular patency. Subsequently, a bleeding point (indicated by a black dotted outline) was created on the MA using a needle. The successful induction of hemorrhage was confirmed by the extravasation of contrast agent (red arrow) (Fig. 7E). Then, 0.2 mL of MNER suspension (with magnetic guidance) and NBCA were injected, respectively. The hemostatic times for MNER and NBCA were 195.0 ± 18.7 s and 225.3 ± 17.9 s, respectively, while the corresponding blood loss was 7.6 ± 0.6 g and 8.3 ± 2.1 g (n = 3; Fig. 7B, C, 7F and 7G). No recanalization was observed in either group, consistent with the permanent embolic effect of both materials. Distal migration was further assessed by quantifying the proportion of embolized vessels. The proportion of embolized vessels was markedly lower in the MNER group (outlined in yellow) than in the NBCA group (outlined in green), indicating that MNER with magnetic guidance provided greater embolization selectivity and reduced the risk of excessive non-target embolization (Fig. 7D).

Fig. 7.

Magnetically controlled targeted embolization of the hemorrhagic small blood vessel with branches. (A) Schematic of targeted embolization in the hemorrhagic small blood vessel with branches. (B, C) Hemostatic time (B) and lost blood (C) in blank, MNER, and NBCA groups. (D) Proportion of non-target vascular embolization in the MNER and NBCA groups. (E) Macroscopic image of the normal, bleeding and embolized small blood vessel. (F–K) DSA images of the normal, bleeding and embolized small blood vessel in MNER with magnetic guidance (MG) (F), NBCA (G), MNER without MG (H), free thrombin (I), Fe₃O₄@SiO₂ without thrombin (J) and non-magnetic thrombin particle (K) groups. Rabbits, n = 3. Data are shown as mean ± SD. Statistical significance was calculated by Student's t-test. ns, not significant, **p < 0.01.

Next, to further elucidate the individual contributions of each component to the in vivo embolization process and clarify the underlying mechanism, the MNER without magnetic guidance, free thrombin, Fe₃O₄@SiO₂ without thrombin, and non-magnetic thrombin particles were included as comparison groups. In the MNER without magnetic guidance, free thrombin and non-magnetic thrombin particle groups, hemostasis failed, likely due to the lack of local aggregation, thrombin was rapidly washed away by blood flow before exerting its coagulation effect (Fig. 7H, I and K). In the Fe₃O₄@SiO₂ without thrombin group, persistent bleeding was observed after embolization, which can be attributed to the inability of thrombin-free particles to form stable fibrin clots (Fig. 7J). These results demonstrated that magnetic aggregation and thrombin activity must act synergistically to achieve controlled and targeted embolization.

3. Conclusion

In this work, we prepared a novel magnetic nano-embolization robot (MNER) with magnetic aggregation and thrombin-triggered coagulation for targeted intravascular embolization. The MNER with core-shell structure was constructed through a sequential four-step synthesis: silica encapsulation of a magnetic Fe₃O₄ core, surface amination, conjugation with a bis-PEG6-NHS crosslinker, and covalent immobilization of thrombin. Upon contacting anticoagulant blood, the thrombin on the surface of the MNER actuated fibrin monomers to connect with each other to form water-insoluble fibrin polymers, which interweaved with each other into a network to trap red blood cells, resulting in forming stable blood clotting. In a static liquid environment, the MNER showed excellent magnetic response movement ability. Moreover, in 3D models of a Y-shaped blood vessel (diameter: 4 mm), an aneurysm (diameter: 6 mm) and a branched blood vessel (diameter: 2 mm) with a circulating fluid environment (flow velocity = 0.2 m/s), the MNER also showed excellent magnetically guided aggregation movement ability. In vivo rabbit experiments confirmed the efficacy and safety of MNER, achieving successful embolization of renal arteries (≈3 mm) without migration or recanalization, as well as targeted occlusion in a hemorrhagic mesenteric artery model with complex branches. Owing to its easy usage, biocompatibility, and effective embolization performance, MNER represents a promising next-generation embolic agent designed for local retention via magnetic aggregation in conjunction with catheter-based delivery.

4. Experimental and methods

Materials: Iron (II, III) oxide (Fe₃O₄), tetraethyl orthosilicate (TEOS), triethanolamine (TEA), fluorescein isothiocyanate (FITC), dimethyl sulfoxide (DMSO), hexadecyltrimethylammonium chloride (CTAC), 1-octadecene, and (3-Aminopropyl) triethoxysilane (APTES) were purchased from Aladdin Chemical Co. Ltd. Bis-PEG₆-NHS ester (M_w = 576.55 g/mol) was purchased from GLPBio Technology Inc. Thrombin was purchased from Shanghai Acmec Biochemical Technology Co., Ltd. Live/dead cell staining kits were purchased from Beyotime (Shanghai, China).

Preparation: First, Fe₃O₄@SiO₂ was synthesized via a biphasic stratification method. 60 mL deionized water (DIW) was heated to 60 °C under constant stirring (500 rpm), followed by the addition of 3 g CTAC and 160 μL TEA to form the aqueous phase. 2.5 mL Fe₃O₄ aqueous suspension (2 wt%) was added into the aqueous phase, after which 20 mL 1-octadecene containing 10 vol% TEOS was slowly added. The reaction proceeded at 60 °C for 15 h. The black products were magnetically separated and washed with ethanol and water. Second, the Fe₃O₄@SiO₂ was dispersed in 50 mL ethanol containing 1 mL APTES and stirred at 25 °C. 12 h later, Fe₃O₄@SiO₂-NH₂ was magnetically collected and washed with ethanol and DMSO for three times. Third, 20 mg Fe₃O₄@SiO₂-NH₂ was added into 10 mL DMSO solution containing 1 mg/mL bis-PEG₆-NHS and 1.7 mM TEA and stirred at 25 °C. 2 h later, Fe₃O₄@SiO₂-NHS was magnetically separated, washed with DMSO and DIW for three times. Finally, Fe₃O₄@SiO₂-NHS suspension was mixed with thrombin aqueous solution (100 U/mL) and stirred at 25 °C (100 rpm). 2 h later, Fe₃O₄@SiO₂-thrombin (MNER) was magnetically separated, resuspended in DIW, and stored at 4 °C. Sometimes, fluorescent thrombin was employed to prepare fluorescent MNER. Prior to the embolization procedure, iohexol was added to the MNER aqueous dispersion and thoroughly mixed to adjust the iohexol concentration to 35 wt%.

Characterization: Fourier-transform infrared (FTIR) spectroscopy (Thermo Scientific Nicolet iS10, USA) and X-ray diffractometer (XRD) were used to characterize Fe₃O₄, Fe₃O₄@SiO₂, Fe₃O₄@SiO₂-NH₂, Fe₃O₄@SiO₂-NHS and MNER. Transmission electron microscope (TEM) and dynamic light scattering (DLS) instrument were used to observe the morphology and size of the MNER. Magnetization loops of the MNER were measured using a vibrating sample magnetometer (VSM) (Lake Shore Cryotronics, Inc., USA, Model 8604 VSM).

Mechanical test: Rheological properties were characterized using an MCR 92 rheometer (Anton Paar) equipped with 25 mm parallel plates at 37 °C. Samples were tested with 1 mm gap height under time-dependent oscillatory sweeps at constant strain (γ = 1%) and angular frequency (ω = 10 rad/s).

Injectability test: The injectability of the material was evaluated using an electronic universal testing machine (MTS, C42.503, USA). The test was conducted by loading the MNER suspension into a 1 mL syringe connected to a microcatheter. The suspension was extruded at a constant flow rate of 1 mL/min, and the corresponding extrusion force was measured in Newtons (N). To ensure reliability, the experiment was performed in quadruplicate (n = 4).

Radiopacity test: For radiopacity quantification, 1 mL syringes containing MNER suspensions containing varying concentrations of iohexol and commercial contrast agent were placed on a CT scanner bed to assess their radiopacity and measure CT values. Furthermore, X-ray fluoroscopy was used to characterize the radiopacity.

Magnetic response test: Magnetic response experiments were performed using a custom-built electromagnetic coil system. The movement velocity of MNER was measured under magnetic fields of 2.5, 8.2, and 15.8 mT generated by an electromagnet. Particle motion was recorded, and the displacement (Δd) of individual particles over a fixed time interval (Δt) was tracked (n = 3). The movement velocity (v) was calculated according to the following formula:

$$v=\Delta d/\Delta t\text{.}$$ (1)

Hemolysis test: For hemolysis testing, rabbit erythrocytes were isolated from whole blood via centrifugation (200 ×g, 10 min), washed with PBS, and resuspended in 0.9% saline to 5% v/v concentration. 100 μL MNER solution was added into 1 mL erythrocyte suspension and incubated at 37 °C. 0.9% saline and DIW were used as negative and positive control groups, respectively. 24 h later, samples were centrifuged (500 ×g, 5 min) and supernatant absorbance (A _{540 nm}) measured. Hemolysis ratio (%) was calculated as:

$$\text{Hemolysis}\text{ratio}(\%)=\left(A_{\mathrm{s}}–A_{\mathrm{p}}\right)/\left(A_{\mathrm{d}}–A_{\mathrm{p}}\right)\times 100\%$$ (2)

A_s, A_p, and A_d were the absorbance of samples, physiological saline, and DIW, respectively. Three independent experiments were performed.

Biocompatibility test: First, 100 μL MNER suspension was added to cultures of human umbilical vein endothelial cells (HUVECs, 2.5 × 10⁴ cells per well) and co-cultured at 37 °C. 24 h later, the HUVECs were stained with live/dead cell staining and assessed using a CCK-8 assay. Furthermore, 12 male Wistar rats (6–8 weeks) were randomly divided into four groups: blank control, 7-day, 14-day, and 28-day (n = 3). Subsequently, the MNER suspension was injected into the dorsal subcutaneous tissue to evaluate its in vivo biocompatibility. The in vivo biocompatibility of the MNER was tested by injecting it into the dorsal subcutaneous tissue of male Wistar rats (6–8 weeks). At a pre-set time, the whole blood was collected for routine blood and serum biochemistry tests. Tissue specimens adjacent to the MNER implants were harvested, followed by fixation, dehydration, and paraffin embedding for subsequent histological analysis.

Renal artery embolization in a rabbit model: A total of 12 male New Zealand rabbits were randomly divided into four groups: blank control, 7-day, 14-day, and 28-day (n = 3). First, anesthesia was induced by intramuscular injection of a combination of Zoletil 50 (5 mg/kg) and 2% xylazine (2 mg/kg). Under the guidance of X-ray fluoroscopy from a digital subtraction angiography (DSA) system (Allura Xper FD20, Philips, Netherlands), a 4-F catheter (Cordis, China) was introduced through a femoral artery incision and navigated to the right renal artery via the abdominal aorta. The patency of the renal artery was verified using an iohexol bolus. Following this, 0.1–0.2 mL of MNER solution was administered slowly into the renal artery for embolization, assisted by an external magnetic field. Post-operative monitoring at days 7, 14, and 28 involved CT, color Doppler flow imaging (CDFI), and contrast-enhanced ultrasound (CEUS) with SonoVue contrast agent. Finally, the rabbits were euthanized, and the target renal arteries along with other vital organs were harvested, fixed in phosphate-buffered formalin, and processed through dehydration and paraffin embedding for subsequent analysis.

Ruptured mesenteric artery embolization in a rabbit model: A total of 18 male New Zealand rabbits were randomly divided into six groups: MNER with magnetic guidance (MG), NBCA, MNER without MG, free thrombin, Fe₃O₄@SiO₂ without thrombin and non-magnetic thrombin particle groups (n = 3). A 2.2-F microcatheter was introduced to the superior mesenteric artery (SMA) under DSA guidance. Active bleeding was induced by puncturing SMA with a 20-G needle. Immediately afterwards, embolization was performed using the following six groups: MNER with magnetic field, NBCA, MNER without magnetic field, free thrombin, Fe₃O₄@SiO₂ without thrombin, and non-magnetic thrombin particles.

Ear central artery embolization in a rabbit model: A total of 3 male New Zealand rabbits were used in this experiment. A 24-G IV catheter was introduced to the ear central artery under DSA guidance. Immediately afterwards, MNER suspension (0.2 mL) was injected into the artery with the aid of a magnetic field.

Summary of Experimental Groups and Sample Sizes for New Zealand Rabbits: A total of 33 New Zealand rabbits were included in this study, with 12 used for renal artery embolization (4 groups, n = 3), 18 for the superior mesenteric artery embolization (6 groups, n = 3), and 3 for ear-central artery embolization. Animals were randomly allocated to experimental groups using a computer-generated random number sequence. The animal use and experimental protocols were reviewed and approved by the Animal Ethics Committee of Zhuhai Bestest Biotechnology Co., Ltd. [IAC25W039(5)].

Statistical analysis: All statistical analyses were conducted using GraphPad Prism software (version 9). Quantitative data are expressed as mean ± standard deviation (SD). For comparisons between two groups, Student's t-test was applied, whereas differences among multiple groups were evaluated by one-way analysis of variance (ANOVA), followed by Tukey's test for post-hoc analysis to determine specific intergroup significances. Statistical significance levels are denoted as ns (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. All experiments were performed with at least three replicates (n ≥ 3).

CRediT authorship contribution statement

Jinming Fan: Writing – original draft, Methodology, Investigation, Conceptualization. Feile Ye: Writing – original draft, Methodology, Data curation. Jiayuan Lin: Methodology, Investigation. Honggang Shi: Methodology. Yifei Hu: Formal analysis. Jincheng Liu: Visualization. Yajun Yuan: Investigation. Xianglong Zhong: Methodology. Yuhan Jiang: Methodology. Shengchao Zhao: Methodology. Jinyan Gu: Methodology. Menghui Liu: Methodology. Qing Zhao: Writing – review & editing. Ke Zhang: Writing – review & editing, Writing – original draft, Visualization, Supervision, Funding acquisition. Xin Peng: Writing – review & editing, Writing – original draft, Supervision, Funding acquisition.

Ethics approval and consent to participate

This study's animal testing protocol received approval from the Animal Ethics Committee of Zhuhai Bestest Biotechnology Co., Ltd. [IAC25W039(5)], and the Animal Ethics Committee of the Fifth Affiliated Hospital of Sun Yat-sen University (00621).

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is/are the supplementary data to this article:

Data availability

Data will be made available on request.