Blood cell membrane camouflaged probiotic-powered micro-robot for dual-targeted thrombus therapy

Abstract

Conventional thrombolytic agents exhibit significant limitations in treating thrombotic disorders characterized by narrow therapeutic windows and persistent vascular injury, failing to meet clinical requirements in terms of precision, timeliness, and treatment durability. To address these challenges, we have developed a blood cell membrane-camouflaged probiotic-powered micro-robot system (EcN-PN/rt-PA). This innovative system utilizes natural probiotics as biological carriers functionalized with drug-loaded hybrid membrane nanovesicles possessing dual-targeting capabilities against thrombus. By harnessing the rapid motility of probiotics and the active targeting properties of the hybrid membranes, the system achieves rapid and precise delivery of thrombolytic drugs to thrombotic tissues. Capitalizing on the innate rod-shaped structure and self-propelling capacity of probiotics, this micro-robot demonstrates enhanced penetration depth into thrombotic tissues, significantly improving thrombolytic efficiency. Experimental results reveal that this thrombolytic micro-robot, which combines thrombus targeting, deep penetration, and prolonged circulation characteristics, not only rapidly and accurately targets and penetrates thrombotic tissues but also extends the biological half-life of thrombolytic drugs approximately 193-fold, enabling sustained thrombolytic effects while effectively inhibiting secondary thrombus formation. This novel dual-targeting thrombolytic strategy provides an important scientific solution for treating thrombotic diseases with narrow therapeutic windows caused by vascular injury.

Graphical abstract

Thrombotic disorders with narrow therapeutic windows remain a critical clinical challenge due to the limitations of conventional thrombolytic agents. Herein, we develop a blood cell membrane-camouflaged probiotic-powered micro-robot (EcN-PN/rt-PA) for targeted thrombolysis. This system uniquely integrates the rapid motility of probiotics with the dual-targeting capability of hybrid membranes, enabling deep penetration into thrombotic tissues and enhanced drug accumulation. Notably, it extends the biological half-life of rt-PA by approximately 193-fold, achieving sustained thrombolytic efficacy and inhibiting secondary thrombosis. This work provides a promising micro-robotic platform for treating vascular injury-induced thrombotic diseases.

1. Introduction

Thrombus, as the critical pathological basis of life-threatening cardiovascular and cerebrovascular diseases, has attracted increasing attention [1,2]. Although thrombolytic drugs remain the primary clinical treatment for thrombotic disorders, their short half-lives necessitate frequent administration and significantly increase the risk of severe hemorrhagic complications such as intracranial hemorrhage [3,4]. Breakthroughs in drug delivery technology have substantially reduced bleeding risks by enhancing thrombolytic drug targeting and prolonging circulatory half-life, thereby effectively preventing hemorrhagic complications associated with systemic high-dose administration [5,6]. A biomimetic drug delivery system inspired by nature has become a cutting-edge therapeutic strategy [7,8]. The biomimetic carrier camouflaged with a cell membrane exhibits advantages over traditional systems, including excellent biocompatibility and the ability to preserve the complex functions of donor cells under physiological and pathological conditions [9,10].

Research has shown that under pathological conditions, platelets and neutrophils synergistically participate in thrombus formation [11,12]. Platelets adhere to the site of vascular injury through surface receptor-mediated mechanisms, activate intracellular signaling pathways, and secrete pro thrombotic factors, promoting the formation of fibrin networks [13]. Thrombosis further triggers an inflammatory cascade reaction, inducing neutrophil chemotaxis towards the thrombus area, followed by the release of neutrophil extracellular traps leading to coagulation dysfunction, while providing structural support for thrombus formation [14]. This has prompted researchers to develop various biomimetic drug delivery strategies based on these pathophysiological mechanisms [15]. This includes platelet membrane encapsulation technology that significantly prolongs the circulation time of thrombolytic drugs [16], as well as our team's previously developed enzyme driven neutrophil drug delivery system [17], which confirms the significant thrombus targeting ability of neutrophils. Therefore, platelet and neutrophil membrane camouflaged thrombolytic drug delivery systems can evade immune clearance and prolong drug half-life, and specifically target thrombus sites.

Cell membrane camouflage drug carriers mainly rely on chemotactic factor gradients, hemodynamics, and passive diffusion mechanisms to enrich towards the lesion site [18,19]. This passive targeting strategy is often limited by delivery efficiency and timeliness [20]. In order to improve the active motion performance of drug carriers, biological micro motor systems have become a research hotspot both domestically and internationally in recent years [21,22]. Their significant advantages are reflected in: enhancing the targeted migration rate [23], enhancing the penetration ability of lesions and optimizing drug release kinetics [24,25]. In the study of the structure-activity relationship of antithrombotic micro-motors, geometric morphology is a key parameter that regulates the lateral migration behavior of blood vessel walls [26]. The currently developed micro-motors mainly include three topological structures: spherical, tubular, and rod-shaped [27,28]. Among them, non-spherical configurations exhibit unique fluid dynamics characteristics [29]. Compared to the laminar motion of spherical particles, the torsional rotational composite motion generated by asymmetric structures during circulation can significantly enhance the interaction with vascular wall thrombosis [30]. More importantly, non-spherical particles have two biological advantages: reduced macrophage clearance rate and prolonged blood circulation half-life, thereby significantly increasing drug accumulation concentration at the thrombus site [5,31].

Despite these advances, existing synthetic micro-motors (e.g., tubular and rod-shaped motors) face critical limitations that hinder their clinical translation for thrombolysis [32,33]. First, most synthetic micro-motors rely on toxic fuels (e.g., H₂O₂) or external fields (e.g., magnetic, acoustic) for propulsion, which raises biocompatibility concerns and limits their autonomous navigation in complex vascular networks [32,34]. Second, conventional nano-carriers such as liposomes, polymeric nanoparticles, and inorganic nanoparticles lack self-propulsion capability and primarily depend on passive diffusion or hemodynamic forces for transport, resulting in poor thrombus penetration and insufficient drug accumulation at the core of the thrombus [35,36]. Third, even non-spherical synthetic carriers, although showing improved circulation, cannot generate continuous autonomous movement, and their surface functionalization often compromises targeting ligand activity [37,38]. These technical gaps underscore an urgent need for a biocompatible, self-propelled, and actively targetable drug delivery platform that can penetrate deep into the thrombus.

Nature offers a compelling solution through non-spherical biological micro-motor systems with autonomous movement ability, mainly including life units such as probiotics, sperm, and muscle cells, which achieve efficient movement through inherent biological behavior [39,40]. Among them, probiotics have been developed as gene therapy carriers, protein delivery platforms, and small molecule drug delivery systems due to their high natural abundance and strong reproductive ability [41,42]. However, most probiotic strains have potential pathogenicity, severely limiting their clinical application [43]. Escherichia coli Nissle 1917 (EcN) is a facultative anaerobic probiotic that is easily cleared by serum in the body and does not have pathogenicity [44]. EcN has natural flagella that can drive its own movement through flagella oscillation, making it a good new type of biological micro motor drug delivery carrier. How to engineer and improve EcN to carry thrombolytic drugs while retaining its own in vivo mobility, and to have active thrombus targeting and deep tissue penetration capability remains a challenge.

Herein, this study proposed a blood cell membrane-camouflaged dual-targeting micro-robot system powered by engineered probiotics for thrombus-targeted and penetrative therapy. As illustrated in Scheme 1, we have developed a probiotic-based thrombolytic micro-robot EcN-PN/rt-PA by utilizing engineered EcN as both a propulsion motor and drug carrier, where the probiotic surface is functionalized with hybrid nanovesicles (PN) composed of platelet membranes (PLM) and neutrophil membranes (NEM) to confer dual thrombus-targeting capability while encapsulating the thrombolytic drug alteplase (rt-PA). This drug delivery system harnesses the flagellar propulsion of probiotics for autonomous motility and leverages their inherent rod-shaped morphology for enhanced tissue penetration. During circulation, the probiotic-powered dual-targeting system exhibits self-navigating capability toward thrombotic sites, effectively penetrating thrombi to achieve local drug accumulation, thereby enabling rapid dissolution of thrombi while minimizing bleeding side effects associated with systemic thrombolysis. The development of this innovative probiotic-driven dual-targeting micro-robot system establishes a promising therapeutic paradigm for clinical management of thrombotic disorders.

Scheme 1.

Preparation and thrombolytic mechanism of blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. (a) Preparation of thrombotic dual targeted nanovesicles PN/rt-PA. (b) Preparation and dual targeted thrombolytic mechanism of blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. Some of the material images are from BioGDP.com

2. Results and discussion

2.1. Preparation and characterization of neutrophil membranes (NEM) and platelet membranes (PLM)

The preparation methods of neutrophil membranes (NEM) are shown in Fig. 1a: Firstly, fresh peripheral blood was extracted from the rats. Then, neutrophils were extracted from peripheral blood using a neutrophil extraction kit. Finally, the neutrophil membrane was separated through three repeated freeze-thaw cycles. The morphological images of extracted neutrophils stained with Wright-Giemsa as shown in Fig. 1b, display the typical “lobulated nucleus” structure of neutrophils. Fig. 1c shows that the extracted neutrophils are elliptical in shape and have many folds on the surface. Flow cytometry was used to analyze neutrophils labeled with specific antibodies against Ly6G and CD11b. The viability and the purity of the neutrophils was 99.9%, as shown in Fig. 1d. The extracted neutrophils were stained with a red lysosome probe and observed by laser confocal microscopy. As shown in Fig. 1e, neutrophils contain a large number of lysosomes and exhibit a typical multinucleated structure. The above results demonstrate that the neutrophil extraction method used in this study is reliable. The neutrophil membrane was extracted by repeated freeze-thaw cycles and analyzed by TEM and SEM. The results are shown in Fig. 1f, and the neutrophil membrane was successfully separated. Western blot experiment was used to detect neutrophil-specific markers (CD11b, Ly6G) in native neutrophils and purified NEM. The results (Fig. S1) showed that neutrophil membranes also highly expressed natural neutrophil specific proteins, proving high membrane purity and the reliability of neutrophil membrane extraction and separation methods.

Fig. 1.

Extraction and characterization of neutrophil membrane (NEM). (a) Extraction method of neutrophil membrane. (b) Morphological images of neutrophil stained with Wright-Giemsa. (c) TEM image of the neutrophil. (d) Flow cytometric analysis of the purity of neutrophil doubly stained with CD11b and Ly6G antibodies. (e) Fluorescent pictures of Lyso-tracker Red and DAPI dye labeled neutrophils. (f) TEM and SEM images of the neutrophil membrane. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

The preparation methods of platelet membranes (PLM) are shown in Fig. 2a: Firstly, fresh peripheral blood was extracted from the rats. Then, platelets were extracted from peripheral blood using a platelet extraction kit. Finally, the platelet membrane was separated through three repeated freeze-thaw cycles. TEM and SEM images of extracted platelet as shown in Fig. 2b, that mostly platelet cells exhibited an oval shaped appearance with multiple pseudopodia. The extracted platelets were stained using cell membrane staining probe DiR and observed using laser confocal microscopy. The results (Fig. 2c) showed that platelet cells exhibited a nuclear free structure. After labelling with platelet specific antibodies CD41 and CD61, flow cytometry analysis was performed. The results are shown in Fig. 2d, the extracted platelet cells expressed a large amount of CD41 and CD61 proteins, proving that the purity of the extracted platelets was high. Fig. 2e shows quantitative immunofluorescence results for CD41 and CD61 in purified platelets. The protein level of CD41 on the platelet membrane was markedly higher than that of CD61. This finding was consistent with the data in Fig. S2, confirming the successful isolation of highly pure platelets and platelet membranes. The platelet membrane was extracted by repeated freeze-thaw cycles and analyzed by SEM (Fig. 2f), TEM (Fig. 2g) and DLS (Fig. 2h). The results proved that the platelet membrane was successfully separated.

Fig. 2.

Extraction and characterization of platelet membrane (PLM). (a) Extraction method of platelet membrane. (b) TEM and SEM images of extracted platelet. (c) Fluorescently labeled platelets with probe DiR analyzed by a laser scanning confocal microscopy system. (d, e) Immunofluorescence analysis of the characteristic proteins (CD61 and CD41) express on the platelet surface. (n = 3; mean ± SD). ∗∗∗P < 0.001. (f, g) TEM and SEM images of platelet membranes extracted from platelets. (h) Particle size of platelet membranes.

2.2. Preparation and characterization of EcN-PN/rt-PA

The synthesis steps of EcN-PN/rt-PA are shown in Fig. 3a: Firstly, the obtained neutrophil membrane and platelet membrane are hybridized, and then internally loaded with rt-PA to obtain thrombus dual targeted hybrid membrane nanovesicles PN/rt-PA. Finally, the drug loaded hybrid nanovesicles were modified onto the surface of probiotics to obtain the blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. The TEM and SEM of the thrombus dual targeted hybrid membrane nanovesicles PN/rt-PA as shown in Fig. 3b,c,d, the particle size of nanovesicles is about 105.71 ± 2.8 nm, exhibiting a distinct shell core structure. The zeta potential results of PN/rt-PA nanovesicles are shown in Fig. S3 was −11.32 mv. The rt-PA protein appears as electron-dense black dots under TEM because proteins have higher electron density than the lipid membrane bilayer, leading to stronger electron absorption and darker contrast. Minor variation in rt-PA distribution is typical for protein loading into membrane vesicles, that slight heterogeneity in vesicle size during extrusion. The encapsulation efficiency and drug loading capacity of rt-PA in PN/rt-PA were quantitatively analyzed using fluorescence quantitation methods (Fig. S4). The rt-PA proteins were labeled with Cy5.5-NHS to obtain Cy5.5-rt-PA, and compared the fluorescence intensity before and after preparation of PN/rt-PA. The drug entrapment efficiency of rt-PA in PN/rt-PA was 87.27%, and the drug loading capacity was 28.52%. The driving forces for rt-PA encapsulation primarily arise from hydrophobic and electrostatic interactions between rt-PA and the phospholipid bilayer of the hybrid membranes. During sonication, membrane fluidization induces transient openings in the lipid bilayer, allowing rt-PA to enter. Subsequent extrusion promotes membrane reconstitution and resealing, thereby trapping the protein inside the vesicles. The above results demonstrate the successful preparation of thrombus dual targeted hybrid membrane nanovesicles PN/rt-PA.

Fig. 3.

Preparation and characterization of blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. (a) The preparation route of the blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. (b, c) TEM and SEM images of thrombotic dual targeted nanovesicles PN/rt-PA. The yellow arrow refers to the neutrophil and platelet membrane hybrid membrane. The white arrow refers to rt-PA. (h) Particle size of thrombotic dual targeted nanovesicles PN/rt-PA. (e) TEM and SEM images of EcN. (f) Analysis of EcN labeled with active probiotic marker dye SYTO9 by laser confocal system. (g) TEM and SEM images of blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. The yellow arrow refers to thrombotic dual targeted nanovesicles PN/rt-PA. (h) A laser confocal fluorescence imaging system was used to analyze micro-robot EcN-PN/rt-PA. Green fluorescence refers to probiotic labeled with probe SYTO9. Red fluorescence refers to rt-PA labeled with probe Cy5.5. Blue fluorescence refers to neutrophil and platelet membrane hybrid membrane labeled with probe DiR. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

A large amount of highly active probiotics EcN were obtained through liquid culture method. The SEM and TEM images of EcN as shown in Fig. 3e, the obtained EcN have good morphology, with a rod length of approximately 2-3 μm. The live bacteria staining fluorescent probe SYTO9 was used to stain EcN and observed through laser confocal microscopy, as shown in Fig. 3f, EcN have a high survival rate and exhibit a significant rod-shaped structure. The PN/rt-PA nanovesicles were modified onto the surface of EcN to obtain the micro-robot EcN-PN/rt-PA. The TEM and SEM images of EcN-PN/rt-PA were shown in Fig. 3g, EcN surface modified with many nanovesicles, and the morphology of probiotics is good. To demonstrate the successful surface modification of EcN with PN/rt-PA, the prepared EcN-PN/rt-PA was observed by fluorescence imaging method. Rt-PA proteins were labeled with Cy5.5-NHS to obtain Cy5.5-rt-PA, and the hybrid cell membrane were labeled with DiR to obtain DiR-PN, then EcN were labeled with SYTO9. Finally, the fluorescent labeled EcN-PN/rt-PA was obtained and observed by laser scanning confocal microscopy. As shown in Fig. 3h–S5, the surface of the green fluorescent probiotic cells had abundant blue and red fluorescence, indicating that the PN/rt-PA successfully modified on the surface of EcN.

The drug entrapment efficiency (EE) and drug loading capacity (DL) of rt-PA in EcN-PN/rt-PA were assessed using fluorescence analysis. Cy5.5-rt-PA and DiR-PN were prepared to obtain fluorescent labeled EcN-PN/rt-PA, and EcN was stained with the probe SYTO9. As shown in Fig. S6, the EE of rt-PA in EcN-PN/rt-PA was relatively high about 65.73%. The corresponding DL of rt-PA in EcN-PN/rt-PA was about 21.71%. These results indicated that the methods used to modify PN/rt-PA on the surface of EcN are reasonable and successful. The retention activity of rt-PA in EcN-PN/rt-PA was compared with the activity of the corresponding concentration of free rt-PA using the chromogenic subtract S-2288. As shown in Fig. S7, the retention activity of rt-PA in PN/rt-PA is 81%, and the retention activity of rt-PA in EcN-PN/rt-PA is 68%, which may be due to the damage caused by the process of modifying EcN with PN/rt-PA. Sodium dodecyl sulfate Polyacrylamide gel electrophoresis (SDS-PAGE) was used to separate and detect the proteins in different components (rt-PA, EcN, PN/rt-PA, EcN-PN/rt-PA and Marker)), as shown in Fig. S8, the result also proved that PN/rt-PA had been successfully modified on the surface of EcN. The above results fully demonstrate the successful preparation of EcN-PN/rt-PA, proving that the synthetic experimental method used is correct.

2.3. Functional characterization of EcN-PN/rt-PA

A live/dead bacteria double stain kit was used to analyze the probiotic survival rate of prepared micro-rod robot EcN-PN/rt-PA. As shown in Fig. 4a and b, confocal fluorescence imaging system was used to analyze the survival rate of EcN-PN/rt-PA. The red fluorescence intensity of EcN-PN/rt-PA gradually increased over time, suggesting a gradual rise in probiotic mortality. Flow cytometry was used to analyze the dead rate of EcN-PN/rt-PA. As shown in Fig. 4c, the bacterial viability of EcN-PN/rt-PA was high. The OD_600nm measurement of EcN and EcN-PN/rt-PA (Fig. 4d) showed that the concentration of probiotic solution EcN-PN/rt-PA was significantly lower than that of EcN over time, this may be due to the modification of PN/rt-PA on the surface of EcN hindering the reproduction of EcN. The agar plate colony counting assays was used to directly quantify EcN-PN/rt-PA viability and proliferation as Fig. S9. The plating results were consistent with OD600nm and confocal data. PN/rt-PA modification inhibited EcN proliferation but maintained high viability, which improves biosafety in vivo. The above results indicate that the survival rate of probiotics after engineering transformation is high, although it affects their reproduction, thereby hindering their proliferation in the body and ensuring high biological safety.

Fig. 4.

Function characterization of blood cell membrane-camouflaged probiotic micro-robot EcN-PN/rt-PA. (a, b) A laser confocal fluorescence imaging system was used to analyze the survival and dead rates of EcN-PN/rt-PA in LB broth medium. Probe SYTO9 labels live probiotic, and probe PI labels dead probiotic. Scale bars, 100 μm. (n = 3; mean ± SD). ∗∗∗P < 0.001. (c) Flow cytometric analysis of the survival and dead rates of EcN and EcN-PN/rt-PA in LB broth medium. Probe SYTO9 labels live probiotic, and probe PI labels dead probiotic. (d) The OD_{600 nm} values of EcN and EcN-PN/rt-PA at different time points (0 h, 1 h, 2 h, 4 h, 6 h). (n = 3; mean ± SD). ∗∗∗P < 0.001. (e) Cumulative release of rt-PA from rt-PA and EcN-PN/rt-PA micro-robot in the medium with or without PLA2. (n = 3; mean ± SD). ∗∗∗P < 0.001. (f) Motion behavior of EcN and EcN-PN/rt-PA in 10% FBS medium: Optical tracking trajectories (20 s). (g) MSD of EcN and EcN-PN/rt-PA in 10% FBS medium. (n = 3; mean ± SD). ∗P < 0.05. (h) The average migration speed of EcN and EcN-PN/rt-PA in 10% FBS medium. (n = 3; mean ± SD). ∗P < 0.05. (i) Deff of EcN and EcN-PN/rt-PA in 10% FBS medium (n = 3). (n = 3; mean ± SD). ∗P < 0.05. (j) The 3D fluorescence images of EcN and EcN-PN/rt-PA labeled with probe SYTO9, analyzed by confocal z-axle layer scanning mode. Scale bars, 100 μm. (k) Fluorescence imaging of EcN and EcN-PN/rt-PA targeting platelets and neutrophils for 3 h. Probe SYTO9 labels live probiotics, probe DiR labels platelets, and probe DAPI labels neutrophils.

Apolipoprotein-associated phospholipase A2 (PLA2), also known as platelet activating factor acetyl hydrolase (PAF-AH), can hydrolyze platelet membranes [45,46], which are highly expressed at the site of thrombus [47,48]. Therefore, PLA2 responsive drug release was used to evaluate the drug release ability of EcN-PN/rt-PA at the thrombus site. To investigate the drug release of rt-PA from EcN-PN/rt-PA, a human rt-PA ELISA kit was used to analyze the content changes of rt-PA release from EcN-PN/rt-PA micro-robot under different media conditions (with or without PLA2, 200 ng/mL). As shown in Fig. 4e, under the condition of the presence of PLA2 (200 ng/mL), the release dose of EcN-PN/rt-PA significantly increased compared to that under the condition without PLA2 as times went on. About 61% of the rt-PA drugs were released from EcN-PN/rt-PA within 48 h, indicating that the drug release of EcN-PN/rt-PA was clearly responsive to PLA2. The above results demonstrate that the EcN-PN/rt-PA system has thrombotic tissue responsive drug release properties.

This study constructs a micro-rod-shaped robot EcN-PN/rt-PA for targeted treatment of thrombus by modifying natural rod-shaped probiotics. Rod shaped probiotics with flagella motors exhibit rapid self-driven movement. In order to investigate the changes in the motility of EcN modified with PN/rt-PA, a 100× objective magnification laser confocal microscope was used to observe and record the motion behaviors of EcN-PN/rt-PA micro-robot. The motion trajectory and corresponding coordinate of EcN and EcN-PN/rt-PA in Movie S1 were analyzed by Fiji software shown in Fig. 4f, revealing typical run-and-tumble zigzag motion derived from flagellar propulsion. The length of the motion trajectory of EcN-PN/rt-PA was similar to EcN within the same period. The mean-squared displacement (MSD), effective diffusion coefficient (Deff) and average migration velocity of particles were analyzed by MATLAB software refer to previous studies [17,49]. As shown in Fig. 4g, h, i, the MSD, velocity and Deff of EcN-PN/rt-PA were all smaller than those of EcN, which might due to the modification of EcN by PN/rt-PA disrupting some flagella structures on the surface of EcN and reducing its mobility. However, EcN-PN/rt-PA still retained most of its motility, and its micro-rod structure was conducive to penetrate thrombus.

To investigate the effect of different concentrations of PN/rt-PA on EcN motility, we set 3 PN/rt-PA concentrations (low: 0.5 mg/mL; medium: 1.0 mg/mL; high: 2.0 mg/mL) to modify EcN, then recorded motion trajectories and calculated velocity, MSD, and Deff. Results (Fig. S10) confirmed that low/medium concentrations only slightly reduced EcN motility, while high concentrations caused a more obvious decrease. Our working concentration (1.0 mg/mL) balances modification efficiency and motility retention, which is reasonable and optimal. The 3D fluorescence images of EcN and EcN-PN/rt-PA labeled with probe SYTO9 were analyzed by confocal z-axis layer scanning mode as shown in Fig. 4j and Movie S2, indicating that within the same scanning plane, the motion ranges of EcN and EcN-PN/rt-PA were similar.

To investigate the in vitro targeting ability of EcN-PN/rt-PA, transwell chambers were used. The experimental diagram was shown in Fig. 4k. Purified neutrophils and platelet cells were cultured at the bottom of the culture dishes, and different groups were added to the upper chamber with a pore size of 0.45 μm and incubated for 3 h. The live EcN was labeled with the probe SYTO9, the neutrophils were labeled with DAPI, and platelets with the probe DiR. The results showed that the quantity of probiotics reached the lower chamber were different in EcN and EcN-PN/rt-PA groups. The platelets and neutrophils target ability of EcN was similar and significantly lower than EcN-PN/rt-PA (Fig. S11), demonstrated that EcN-PN/rt-PA micro-robot has significant platelet and neutrophils targeting ability. We added a static incubation control (placed on ice, 0 °C) to inhibit the active movement of EcN and EcN-PN/rt-PA. The migration of probiotics in the 37 °C active motion group and 0 °C static inhibition group was compared. The results (Fig. S12) confirmed that the migration of EcN-PN/rt-PA to platelets/neutrophils was significantly higher at 37 °C than at 0 °C, confirming the targeting effect mainly comes from active movement and specific recognition, not passive diffusion. The above results demonstrate that EcN-PN/rt-PA micro-robot has significant self-movement ability and the ability to actively target thrombus establishing the foundation for the in vivo thrombus targeting movement of robot.

2.4. In vitro thrombolysis efficacy and thrombus penetration evaluation of EcN-PN/rt-PA

To investigate the in vitro thrombolytic effect of EcN-PN/rt-PA micro-robot, the method shown in Fig. 5a was used: Firstly, fresh aortic blood was extracted from the rat and equal volumes were placed into a syringe tube; After the blood naturally coagulated, different thrombolysis groups ((1) PBS, (2) rt-PA, (3) PN/rt-PA, (4) EcN-PN/rt-PA, 1 × 10⁵ CFU/mL) were added; At different time points, the dissolved blood was aspirated and added to a blank syringe tube, and the volume of dissolved blood at different time points will be measured. Additionally, thrombus blocks containing thrombolytic agents were removed at different time points for frozen section analysis. After 6 h of incubation at room temperature, solid-liquid separation was performed once per hour, and the volume of liquid blood obtained from thrombolysis was recorded. The thrombolysis results shown in Fig. 5b and c, the thrombolysis volume in each group showed an increasing trend over time. At 6 h, the thrombolysis volume of rt-PA and PN/rt-PA was significantly higher than that of the control PBS. The thrombolysis volumes of EcN-PN/rt-PA were significantly higher than that of rt-PA and PN/rt-PA, which could be attributed to the self-propelling ability of EcN and its small size with natural rod-shaped structure, which facilitated penetration into thrombus tissue, thereby enhancing the thrombolytic effect of rt-PA from EcN-PN/rt-PA.

Fig. 5.

In vitro thrombolysis and thrombus penetration evaluation of EcN-PN/rt-PA micro-robot. (a) Experimental steps for in vitro thrombolysis.Some of the material images in the picture are from BioGDP.com (b, c) Thrombolysis efficacy evaluation of different groups ((1): PBS, (2): rt-PA, (3): PN/rt-PA, (4): EcN-PN/rt-PA) for 6 h in vitro. (n = 3; mean ± SD). ∗∗∗P < 0.001. (d, e) The fluorescence microscopy analysis of the penetration ability of EcN and EcN-PN/rt-PA into blood clots at various time points. Probe SYTO9 labels live probiotic, and probe DAPI labels cell nucleus. (n = 3; mean ± SD). ∗∗∗P < 0.001. (f) Hemolysis rate of red blood cells after incubated with different samples. (n = 3; mean ± SD). (g) Pharmacokinetic study of rt-PA and EcN-PN/rt-PA micro-robot over 48 h (n = 5; mean ± SD). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

To evaluate the thrombus penetration ability of EcN-PN/rt-PA micro-robot, EcN was stained with the probe SYTO9, and thrombus blocks from the syringe at different time points were removed, frozen and sliced, and observed under a confocal microscope. As shown in Fig. 5d and e, the depth of penetration of two groups of probiotics into the thrombus block increased with time, the penetration depth of EcN-PN/rt-PA was lower than EcN. This might be due to the complex surface modifications of probiotics interfering with some of their physiological movements, and the thrombus penetration effect of EcN-PN/rt-PA micro-robot primarily relied on the motility of EcN. The labeled strains were incubated with prepared thrombi, and an in vivo fluorescence imaging system was applied to record the dynamic penetration process and penetration depth. As shown in Fig. S13, EcN exhibited a faster and deeper penetration capacity than EcN-PN/rt-PA, which was consistent with the confocal microscopy results. The above results demonstrate the excellent thrombus penetration and thrombolytic effect of the EcN-PN/rt-PA micro-robot. This mechanism is based on the autonomous movement ability of probiotics combined with surface functionalization with thrombolytic agents, which contributes to the thrombolytic effect of EcN-PN/rt-PA during self-propelled motion. Clot dissolution facilitates deeper penetration of EcN-PN/rt-PA micro-robots into thrombus, enhancing localized therapeutic effects. The effective thrombus penetration and thrombolytic activity of the EcN-PN/rt-PA demonstrate the feasibility of this probiotic-based drug delivery strategy for thrombolytic therapy.

The hemolytic abilities of different samples in vitro were tested as shown in Fig. 5f, the hemolysis rates of all the samples except for water were below 5%, indicating the good blood compatibility of the samples. To verify that the EcN-PN/rt-PA micro-robot drug delivery system constructed in this study serves as an effective drug carrier, its ability to significantly prolong the circulation time of rt-PA was assessed. Pharmacokinetic experiments were used to analyze the biological half-life of rt-PA and EcN-PN/rt-PA as shown in Fig. 5g. The drug concentration-time curve showed a second peak, which might be due to the rupture of nano vesicles PN/rt-PA on the surface of EcN, leading to an increase in drug concentration of rt-PA in the blood. The biological half-life of rt-PA was particularly short, at about 0.07 h, while the biological half-life of EcN-PN/rt-PA was significantly prolonged to 13.52 h. The mean retention time of the EcN-PN/rt-PA was 19.51 h, which was almost 177 times that of free rt-PA (Table S1). The above results demonstrate that EcN-PN/rt-PA micro-robot has a long circulation effect in the body, which can significantly improve the biocompatibility of rt-PA, achieve sustained thrombolytic effect, and inhibit the formation of secondary thrombus.

2.5. Construction of in vivo thrombus model and in vivo thrombus targeting of EcN-PN/rt-PA

To evaluate the thrombus targeting and thrombolytic effect of EcN-PN/rt-PA micro-robot in vivo, an in vivo thrombus model was established using the method shown in Fig. 6a. Firstly, the left common carotid artery blood vessel of C57BL/6 mice was surgically exposed. Then, 10% FeCl₃-soaked filter paper was attached to the bifurcation of the left common carotid artery (LCCA) blood vessels for 5 min. After removing the filter paper for 10 min, emboli could be clearly observed at the bifurcation of the blood vessels. The wound was sutured, and pre- and post-administration vascular conditions were analyzed via ultrasound imaging. Small-animal ultrasound imaging was employed to analyze the LCCA before and after thrombus model establishment, utilizing multiple analytical models. As shown in Fig. 6b and c Movie S3 and Movie S4, it could be clearly observed that the blood vessels at the bifurcation of the carotid artery become narrowed after the construction of the thrombus model in B mode, which was also confirmed by M mode. Colour Doppler mode and pulse-wave (PW) mode further indicated that the blood flow at the bifurcation of the LCCA blood vessels significantly decreased and tended to disappear before and after the construction of the thrombus model. We have supplemented hemodynamic monitoring data at 30 min and 60 min after model establishment. Results (Fig. S14) show that blood flow velocity and perfusion remain stably reduced within 1 h after thrombus formation. No spontaneous recanalization or thrombus shedding occurred, proving the model is stable and reliable for subsequent drug efficacy evaluation. H&E and Masson staining analysis was performed on the LCCA blood vessels of the thrombus model and the sham blood vessels. As shown in Fig. 6d, the results confirmed successful thrombus model construction, with vessels nearly filled with emboli.

Fig. 6.

Construction of in vivo thrombus model and evaluation of in vivo thrombus targeting. (a) Construction of left common carotid artery (LCCA) thrombus model. (b) The left common carotid artery thrombus model was constructed and analyzed by multiple ultrasound imaging modalities. The yellow arrows refer to the FeCl₃ patch location. (c) Changes in systolic and diastolic blood velocity of LCCA before and after thrombus model construction. (n = 3; mean ± SD). ∗∗∗P < 0.001. (d) H&E and Masson staining analysis of blood vessels in thrombus model and control group. (e, f) Thrombus targeting analyze of different groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA) at different time points. Probe Cy5.5-NHS labels rt-PA and prepare fluorescent labeled rt-PA, PN/rt-PA and EcN-PN/rt-PA. (n = 3; mean ± SD). ∗∗∗P < 0.001. (g) Fluorescence imaging analysis of LCCA vessels from different groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA) at 60 min. Probe Cy5.5-NHS labels rt-PA and prepare fluorescent labeled rt-PA, PN/rt-PA and EcN-PN/rt-PA. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

To investigate whether EcN-PN/rt-PA micro-robot can target thrombus tissue in vivo, a LCCA thrombus model was constructed. Cy5.5 was used to label the rt-PA, and prepared the fluorescent labeled PN/rt-PA and EcN-PN/rt-PA. Thrombotic model mice were administered different Cy5.5 fluorescent probe labeled samples (rt-PA, PN/rt-PA, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL) through the tail vein and observed by a small animal imaging device for 1 h. The results shown in Fig. 6e and f, there was almost no fluorescence observed at the thrombus site in the group rt-PA of mice within 1 h. However, significant fluorescence observed at the thrombus site in PN/rt-PA and EcN-PN/rt-PA groups of mice, and it could be observed as early as 10 min, proving the thrombus targeting ability of platelet membrane and neutrophil membrane hybrid nanovesicles (PN). The fluorescence intensity of EcN-PN/rt-PA group is significantly higher than that of PN/rt-PA group and gradually increased over time, this result may be due to the EcN has autonomous mobility which accelerates the rt-PA targeted delivery. One hour after treatment, we collected major organs including the carotid artery, heart, liver, spleen, lung and kidney for fluorescence imaging. The results showed in Fig. S15 indicated that rt-PA, PN/rt-PA and EcN-PN/rt-PA primarily accumulated in the liver, lung and kidney. Notably, EcN-PN/rt-PA exhibited markedly higher accumulation in the spleen than the other two groups. This phenomenon is largely attributed to the spleen's inherent physiological function as an immune organ for substance clearance. Additionally, the EcN-PN/rt-PA treatment group showed substantially increased distribution in the carotid artery, which confirms its favorable thrombus-targeting capability.

Slice analysis of LCCA blood vessels was performed in different groups of mice (rt-PA, PN/rt-PA, EcN-PN/rt-PA) 1 h after administration. Cy5.5 was used to label rt-PA, and fluorescently labeled PN/rt-PA and EcN-PN/rt-PA were prepared. The results (Fig. 6g–S16) showed that the fluorescence intensity of group PN/rt-PA and EcN-PN/rt-PA was significantly stronger than rt-PA group, which proved that PN/rt-PA provided thrombus-targeting ability for EcN-PN/rt-PA. The fluorescence intensity of group EcN-PN/rt-PA was significantly stronger than PN/rt-PA group. We quantified CFU counts in the heart, liver, spleen, lung, kidney and carotid artery tissues of model and sham-operated mice 1 h after injection with EcN-PN/rt-PA. As illustrated in Fig. S17, bacterial colonies predominantly accumulated in the spleen, liver and carotid arteries. Notably, the model group exhibited substantially more colonies in the carotid arteries than the sham group, demonstrating the prominent thrombus-targeting capacity of the EcN-PN/rt-PA system. The above results demonstrate that EcN-PN/rt-PA has significant thrombus targeting ability and exhibits significant enrichment in thrombus tissue. The thrombus targeting effect of EcN-PN/rt-PA originates from the modification of hybrid nanovesicles (PN), and the autonomous exercise ability of EcN, which proves the rationality of this study for the design of modifying to EcN for thrombus treatment.

2.6. Thrombolysis efficacy evaluation of EcN-PN/rt-PA in vivo

An in vivo LCCA thrombus model was constructed using the method described in Fig. 7a. After intravenous administration of different treatments, Doppler ultrasound analysis was performed at different time points. Thrombus model mice were randomly divided into three groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL), with a dose of rt-PA of 2 mg/kg. Doppler ultrasound in PW mode was used to analyze the blood flow of LCCA vessels in different treatment groups of mice at different times. As shown in Fig. 7b,c,d,e, three groups all exhibited significant thrombolytic effects. The rt-PA group achieved peak blood flow in approximately 60 min, while the PN/rt-PA group reached it in about 40 min. This difference might be attributed to the thrombotic targeting of hybrid nanovesicles, which enhanced the drug concentration of rt-PA in thrombotic tissue. The thrombolysis efficiency of EcN-PN/rt-PA was faster than that of PN/rt-PA within 30 min, indicating that EcN played an important role in drug delivery of rt-PA. Through the autonomous movement and thrombus-penetrating ability of EcN, the thrombolysis time was shortened.

Fig. 7.

Experimental steps for in vivo thrombolysis evaluation. (b) Doppler ultrasound analysis of blood velocity in different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA) in left common carotid artery thrombus model mice. (n = 3; mean ± SD). Gray-scale bars 0-255. (c, d, e) Analysis of systolic and diastolic blood pressure in different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA). (n = 3; mean ± SD). (f) H&E and Masson staining analysis of blood vessels from different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA). (g) Cell membrane probe DiR staining analysis of blood vessels from different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA).

However, rt-PA group showed blood flow obstruction at 180 min, resulting in secondary thrombus. The sham treatment group (Fig. S18) exhibited negligible thrombus recanalization in the saline control. PN/rt-PA and EcN-PN/rt-PA treatment groups have no secondary recurrence of thrombosis within 210 min, which may due to the increased circulation time of rt-PA in the drug carrier PN, and the drug carriers containing probiotics EcN can significantly increase thrombolysis time compared to drug carriers without probiotics. Laser Speckle Contrast Imaging system was used to quantitatively monitor blood perfusion during thrombolysis in Fig. S19. LSCI showed that blood flow recovery in the EcN-PN/rt-PA group was faster and more stable than in the rt-PA and PN/rt-PA groups. No re-occlusion was observed within 210 min, supporting stronger thrombolysis and anti-secondary thrombosis effects, which was consistent with the Doppler ultrasound results. After 210 min treatment, we separated the left carotid arteries of each experimental group and compared the relative thrombus sizes of different experimental groups, and the results (Fig. S20) also demonstrated that EcN-PN/rt-PA has a significant thrombolytic effect compared to other group.

The results of H&E and Masson staining analysis of blood vessels from different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA) after 210 min of treatment are shown in Fig. 7f–S21. The thrombus therapeutic effect of PN/rt-PA was significant stronger than that of rt-PA. Moreover, the thrombus therapeutic effect of EcN-PN/rt-PA was stronger than PN/rt-PA, proving that the modification of EcN offered the best thrombus therapeutic effect and inhibited the formation of secondary thrombus. As shown in Fig. 7g–S22, cell membrane staining probe DiR staining analysis of vascular sections from different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA) also demonstrated that EcN-PN/rt-PA has a significant therapeutic effect on thrombosis. The above results demonstrate that the EcN-PN/rt-PA micro-robot constructed in this study has ideal thrombolytic therapy effects and significant inhibitory effects on secondary thrombus.

2.7. The biological safety of EcN-PN/rt-PA in vivo

In order to investigate the biological safety of the EcN-PN/rt-PA micro-robot, the hemorrhagic time of mice treated with saline, rt-PA, PN/rt-PA, or EcN-PN/rt-PA (1 × 10⁵ CFU/mL) at a dose of 2 mg/kg rt-PA was analyzed. As shown in Fig. S23, the hemorrhagic time of rt-PA was the longest compared to the other three groups, indicated sever hemorrhagic side effects. The hemorrhagic time of EcN-PN/rt-PA were slightly shorter than that of PN/rt-PA, possibly due to their faster thrombolytic effect, which was attributed to its thrombus targeting properties and autonomous movement, which reducing the accumulation of thrombolytic drugs in non-target tissues, thereby reducing the occurrence of hemorrhagic side effects. These results demonstrate that EcN-PN/rt-PA not only prolongs the biological half-life of rt-PA but also markedly decreases its hemorrhagic complications, validating both the mechanistic rationale and clinical potential of our thrombolytic strategy.

In order to verify the biological safety of administering large quantities of probiotics intravenously into the bloodstream, the blood routine and biochemical results of mice in different treatment groups (saline, EcN, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL) were analyzed. As shown in Fig. S24, the blood routine parameters of the EcN-PN/rt-PA group were comparable to those of the control group, with no significant difference. However, some parameters of EcN showed significant differences compared to the control group, such as LYM#, LYM%, MONO#, MONO%, PLT and WBC decreased, while NEUT% increased. The changes in these parameters indicated an inflammatory response in the mice due to probiotic infection. The biochemical results (Fig. S25) revealed similar trends, with the EcN-PN/rt-PA biochemical parameters closely matching the control group, while EcN-treated mice exhibited elevated levels of ALT, AST, LDH, CK, CK-MB, CREA and UA. The changes in these parameters revealed that mice suffer from hepatitis and nephritis. H&E staining analysis was performed on the main organs of mice in different treatment groups. As shown in Fig. S26, the liver and kidney tissues of the EcN group exhibited inflammatory cell infiltration, showing pathological features of hepatitis and nephritis. However, the EcN-PN/rt-PA treatment group showed no significant damage to any major organs. The above results indicate that unmodified probiotics can exhibit biological toxicity when administered intravenously, but surface modification of probiotics can significantly reduce their biological toxicity, potentially by inhibiting their reproduction.

3. Conclusion

This study innovatively developed a blood cell membrane-camouflaged probiotic-powered micro-robot system that utilizes natural probiotics as biological carriers functionalized with drug-loaded hybrid membrane nanovesicles possessing dual thrombus-targeting capabilities. By integrating the autonomous motility of probiotics with the active targeting properties of blood cell membranes, this system achieves precise drug delivery to thrombotic sites. Specifically: (1) the low-immunogenicity EcN strain serves as the propulsion core, with its natural flagella structure enabling rapid self-motility; (2) the hybrid membrane modification (platelet and neutrophil membranes hybrid) provides dual-targeting functionality for specific thrombus recognition; (3) the inherent rod-shaped bacterial structure demonstrates significantly enhanced thrombus penetration efficiency compared to conventional spherical carriers. In vitro and in vivo studies confirmed the system's ability to: achieve high drug accumulation at thrombus sites, extend thrombolytic drug half-life approximately 193-fold, and effectively inhibit thrombus recurrence. Further investigations revealed that this modification strategy successfully addresses the biocompatibility challenges of unmodified probiotics (evidenced by reduced serum inflammatory markers and normalized hepatic/renal indices), providing critical evidence for clinical translation. This innovative dual-mechanism thrombolytic platform, combining biological propulsion with active targeting, establishes a new paradigm for treating thrombotic disorders with narrow therapeutic windows.

4. Experimental section

4.1. Extraction and characterization of neutrophil membrane (NEM)

Neutrophils were isolated from mouse peripheral blood using a discontinuous density gradient method. Briefly, whole blood was collected from 8-week-old male C57BL/6 mice via eyeball extraction and suspended in 1 mL of RPMI-1640 medium to obtain 2 mL of peripheral blood cell suspension. A 15 mL sterile centrifuge tube was pre-coated with bovine serum (excess removed), followed by sequential layering of 1 mL Histopaque-1119 and 1 mL Histopaque-1077 at room temperature. The 2 mL blood suspension was carefully layered drop wise onto the gradient and centrifuged at 700×g for 45 min (room temperature). After centrifugation, five distinct layers formed, with two visible opaque interfaces. The solution within 0.5 cm of the bottom opaque layer was collected, washed with 10 mL PBS, and centrifuged at 800×g for 5 min. The final pellet contained purified neutrophils after supernatant removal.

Secondly, neutrophil membranes were extracted and purified from the neutrophil suspension. One milliliter of neutrophils suspension was added into a 2 mL cryopreservation tube. The cryopreservation tube was placed in liquid nitrogen for 10 min, then immediately transferred to a 37 °C water bath. After about 15 min, the cryopreservation tube was removed and returned to room temperature. This freeze-thaw process was repeated three times as described above. After the final thaw, the ruptured neutrophils suspension was further centrifuged (2000×g, 5 min) to prepare neutrophils membrane vesicles. The obtained neutrophils membrane vesicles were suspended in 10 mM PBS and washed three times with PBS containing protease inhibitors by centrifugation (2000×g, 5 min) to separate neutrophils membranes and their organelles, resulting in purified neutrophils membranes. TEM, SEM and laser scanning confocal microscopy were used to analyze the morphology of neutrophils and the neutrophils membranes vesicles. Flow cytometry was used to analyze the expression of neutrophils characteristic proteins CD11b and Ly6G.

4.2. Extraction and characterization of platelet membrane (PLM)

Firstly, platelets were extracted and purified from peripheral blood using the following protocol: After anaesthetizing C57BL/6J strain mice with isoflurane, blood was collected from the eyeballs [50,51] and placed in anticoagulant tubes. Two millilitres of fresh anticoagulated blood were diluted with an equal volume of 10 mM PBS. Platelets were separated from the blood using a human peripheral blood platelet separation kit (P6390, Beijing Solarbio Science & Technology Co., Ltd.). Briefly, 4 mL of platelet separation solution was added to a centrifuge tube, followed by 2 mL of 60% separation solution (1200 μL platelet separation solution + 800 μL tissue dilution solution) were added to form a separation interface. Then, 4 mL of diluted anticoagulant whole blood (prepared within 2 h) were added, and the tube was centrifuged at room temperature for 20 min at a speed of 350×g. After centrifugation, the platelet layer was carefully aspirated into a 15 mL clean centrifuge tube, and 10 mL of PBS was added, followed by centrifugation at 500×g for 20 min. After centrifugation, the supernatant was removed, and 1 mL of PBS was added to obtain a platelet suspension.

Secondly, platelet membranes were extracted and purified from the platelet suspension. One milliliter of the platelet suspension was added into a 2 mL cryopreservation tube. The cryopreservation tube was placed in liquid nitrogen for 10 min, then immediately transferred to a 37 °C water bath. After about 15 min, the cryopreservation tube was removed and returned to room temperature. This freeze-thaw process was repeated three times as described above. After the final thaw, the ruptured platelet suspension was further centrifuged (2000×g, 5 min) to prepare platelet membrane vesicles. The obtained platelet membrane vesicles (PMV) were suspended in 10 mM PBS and washed three times with PBS containing protease inhibitors by centrifugation (2000×g, 5 min) to separate platelet membranes and their organelles, resulting in purified platelet membranes. TEM, SEM and laser scanning confocal microscopy were used to analyze the morphology of platelet and platelet membrane vesicles. Laser scanning confocal microscopy was used to analyze the expression of platelet characteristic proteins CD61 and CD41.

4.3. Preparation and characterization of PN/rt-PA

First, hybrid membranes of platelets and neutrophils were prepared using a repeated freeze-thaw method. Specifically, the precipitates of neutrophils (1 × 10⁹/mL) and platelets (1 × 10⁹/mL) were taken, suspended in 2 mL PBS, and thoroughly mixed. The suspension was subjected to rapid freezing in liquid nitrogen (1 min) → rapid thawing at 37 °C (3 min), repeated three times. After the final thawing, the lysed cell suspension was further centrifuged (4 °C, 1500 rpm, 10 min) to collect the supernatant. The precipitate was suspended in PBS and centrifuged again (4 °C, 14000 rpm, 30 min) to obtain mixed cell membrane vesicles. The resulting hybrid membrane vesicles were suspended in PBS and washed three times (2000 g, 5 min) with PBS (containing protease inhibitors) to separate the mixed cell membranes from cellular organelles. The purified hybrid membrane vesicles were then mixed with the thrombolytic drug rt-PA (10:1 w/w), sonicated for 10 min at 20% power, and extruded five times through a 200 nm polycarbonate porous membrane using an Avestin mini-extruder, ultimately yielding drug-loaded hybrid membrane nanovesicles PN/rt-PA.

TEM, SEM and DLS were used to analyze the morphology of PN/rt-PA nanovesicles. The encapsulation efficiency (EE) and drug loading (DL) of rt-PA in PN/rt-PA nanovesicles were analyzed by protein quantification method using a BCA protein assay kit (YEASEN Biotechnology (Shanghai) Co., Ltd). The EE% was calculated as (Amount of rt-PA loaded in PN/rt-PA/Total amount of rt-PA added) ×100%. The DL% was calculated as (Amount of rt-PA loaded in PN/rt-PA/Total weight of SW-rt-PA) ×100%. The amount of protein loaded in PN/rt-PA was calculated by the equation: amount of protein bound = total protein amount - free protein amount. The total weight of PN/rt-PA was determined from the weight of the freeze-dried PN/rt-PA product.

4.4. Preparation and characterization of EcN-PN/rt-PA

Firstly, we cultured large quantities of highly active Escherichia coli Nissle 1917. The EcN was obtained from BeNa Culture Collection (BNCC361741). The bacteria were cultured in Luria-Bertani (LB, BNCC372161) broth and grown to the logarithmic phase with continuous shaking at 220 rpm at 37 °C.

To functionalize the obtained drug-loaded hybrid membrane nanovesicles PN/rt-PA onto the surface of EcN, we employed a biotin-streptavidin binding strategy to prepare EcN-PN/rt-PA. First, the drug-loaded hybrid membrane nanovesicles were reacted with sulfo-NHS-biotin to synthesize (PN/rt-PA)-Biotin: sulfo-NHS-biotin (1 mL, 500 μM in PBS) was mixed with PN/rt-PA solution (1 mL, 1 mg/mL) and stirred in the dark for 30 min. After the reaction, the mixture was dialyzed in PBS using a 2 kDa dialysis bag for 12 h, and the resulting (PN/rt-PA)-Biotin was dispersed in 2 mL PBS and stored at low temperature. Next, EcN (1 × 10⁴ CFU) was sequentially incubated with sulfo-NHS-biotin (1 mM, 500 μL), streptavidin (1 mM, 500 μL), and the synthesized (PN/rt-PA)-Biotin (1 mg, 500 μL) at 37 °C for 1 h each. Between each incubation step, the cells were washed three times with PBS to remove unbound substances. The final product, EcN-PN/rt-PA was stored at 4 °C for subsequent use.

TEM, SEM and laser scanning confocal microscopy were used to analyze the morphology of EcN and the EcN-PN/rt-PA micro-robot. For CLSM analysis of EcN-PN/rt-PA, three different components were stained as follows: Cy5.5-NHS (10 μM) was used to label rt-PA proteins, and the cell membrane staining probe DiR (5 μM, TargetMol, USA; TD0085) was used to stain the hybrid membrane PN, and EcN stained using the live probiotic staining probe SYTO9 (5 μM). Then, fluorescently labeled SYTO9-EcN-(DiR-PN/rt-PA-Cy5.5) with preparation steps as above. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE, P1250M, Beijing LABLEAD Inc.) was used to separate and detect the proteins in different components (rt-PA, EcN, PN/rt-PA, EcN-PN/rt-PA, Marker). The activity of rt-PA was measured with the chromogenic subtract S-2288 (Chromogenix, no. 82085239). The kinetic activity of rt-PA in EcN-PN/rt-PA was determined by mixing EcN-PN/rt-PA PBS solution with 1 mM assay buffer (S-2288 in 0.1 mM Tris-HCl pH 8.4) in a 96-well cell culture plate. Absorbance at 405 nm was recorded, and the kinetic activity η was determined as the linear slope of ΔA = A(t) − A(0) versus test time t, η = ΔAt. The activity retention of EcN-PN/rt-PA was determined as the ratio of its kinetic activity to the kinetic activity of free rt-PA solution at the same concentration.

To evaluate the EE and DL of rt-PA in EcN-PN/rt-PA, rt-PA-Cy5.5, DiR-PN, and SYTO9-EcN were prepared to obtain SYTO9-EcN-(DiR-PN/rt-PA-Cy5.5). Fluorescence spectrophotometry was used to measure the fluorescence intensity of rt-PA-Cy5.5 before and after modification. EE% = (fluorescence intensity of Cy5.5 after modification/fluorescence intensity of Cy5.5 before modification) ×100%. DL% was determined by comparing the fluorescence intensities of Cy5.5 with those of Cy5.5+DiR + SYTO9 in SYTO9-EcN-(DiR-PN/rt-PA-Cy5.5), using the following equation: DL% = (fluorescence intensity of Cy5.5/fluorescence intensity of Cy5.5+DiR + SYTO9) ×100%.

4.5. Probiotic activity analysis of EcN-PN/rt-PA

The probiotic activity of EcN-PN/rt-PA at different time points was analyzed using laser confocal microscopy (Leica SP8, 100×) and flow cytometry. Live and dead bacteria were stained with SYTO9 and PI probes (Live and dead bacteria kit, Shanghai Maokang Biotechnology Co., Ltd., Shanghai, China, MX 4234) in equal proportions, respectively. The excitation and emission wavelengths of SYTO9 and PI were 480 nm/500 nm and 490 nm/635 nm, respectively. The live and dead bacteria of EcN-PN/rt-PA were observed at different times (0 h, 1 h, 2 h, 3 h, 4 h, and 6 h). A UV spectrophotometer was used to measure the OD values of different bacterial solutions (EcN, EcN-PN/rt-PA) at an absorbance wavelength of 600 nm.

4.6. Drug release of rt-PA from EcN-PN/rt-PA

Dialysis method [[52], [53], [54]] was used to investigate the rt-PA drug release from EcN-PN/rt-PA. Briefly, rt-PA (1 mg) and EcN-PN/rt-PA (1 × 10⁵ CFU/mL) were placed in equal proportions into a dialysis bag (100 kDa) and immersed in a solution medium either with or without PLA2 (200 ng/mL). The dialysis bag was then placed in a shaker and shaken at room temperature for 48 h. At predetermined time intervals, 20 μL samples of the different solutions (rt-PA -PLA2, rt-PA + PLA2, EcN-PN/rt-PA-PLA2, EcN-PN/rt-PA + PLA2) were withdrawn from the dialysis bag. A UV spectrophotometer (absorption wavelength: 205 nm) and a human rt-PA ELISA kit (Shanghai Jianglai Biology Co, Ltd. JL14148-48T) were used to analyze the samples at different time points (1 h, 2 h, 4 h, 6 h, 8 h, 10 h, 12 h, 24 h, and 48 h).

4.7. The motion behaviors evaluate of EcN-PN/rt-PA

To investigate the motility of EcN modified with PN/rt-PA, laser confocal microscopy (Leica SP8, 100×) was used to observe the motility of EcN before and after modification. The movement videos of EcN and EcN-PN/rt-PA (1 × 10⁵ CFU/mL) in serum were shown in Movie S1, the motion trajectory and coordinates from video were analyzed by Image J software. MATLAB software with operation the code written by the research group were used to calculate MSD value, motion velocity, and Deff value based on the coordinate data of particle motion trajectory. The MSD was calculated using the following equation, MSD (Δt) = 4 D Δt + v²Δt², where D is the diffusion coefficient, t is the time, and v is the motion velocity [17].

4.8. In vitro target analysis of EcN-PN/rt-PA

To investigate the platelet targeting ability of EcN-PN/rt-PA, a Transwell chamber (0.45 μm) was used. Purified platelet cells labeled with DiR probe (5 μM) and neutrophils labeled by DAPI probe (5 μM) were cultured in the lower chamber. Meanwhile, 0.2 mL of different bacterial solutions (EcN, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL) were added to the upper chamber with equal colony counts. After a 3-h incubation, the upper bacterial solution was aspirated, and the upper chamber was removed. The live bacterial staining probe SYTO9 (5 μM) was used to stain EcN from different groups present in the lower chamber. The samples were then observed under a confocal laser microscope with excitation and emission wavelengths of 480 nm/500 nm (SYTO9), 358 nm/461 nm (DAPI) and 748 nm/780 nm (DiR). The Image J software was used to analyze the fluorescence intensity of the different groups.

4.9. In vitro thrombolysis experiment analysis of EcN-PN/rt-PA

In order to investigate the in vitro thrombolytic effect of EcN-PN/rt-PA micro-robot, an equal volume of fresh blood was loaded into a syringe tube and divided into 4 groups. Each group was treated with 200 μL of the respective sample (PBS, rt-PA, PN/rt-PA, or EcN-PN/rt-PA; 1 × 10⁵ CFU/mL, corresponding to 0.56 mg/mL rt-PA for the rt-PA-containing groups). The samples were left to stand at room temperature for 6 h, with solid-liquid separation performed hourly. The volume of liquid blood obtained from thrombolysis was recorded, and statistical analysis was conducted using GraphPad software (version 9.0).

To evaluate the thrombus penetration ability of EcN-PN/rt-PA micro-robot, EcN was stained with the probe SYTO9 (5 μM). Thrombus blocks from the syringe of different groups of probiotics (EcN, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL) were removed at different time points as previously described. These thrombus blocks were then frozen, cross-sectioned, and observed under a fluorescence microscope (20×) at different time points (1 h and 3 h). The penetration depth of the probiotics in each group was recorded and analyzed using Photoshop software (2022 version) and GraphPad software (version 9.0).

4.10. Pharmacokinetic evaluation of EcN-PN/rt-PA in vivo

SD rats (250 ± 10 g, n = 5) were provided with free access to food and drinking water. A mouse left common carotid artery thrombus model was constructed as above, and the rats were randomly divided into two groups (rt-PA, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL, corresponding to 0.56 mg/mL rt-PA) freely. Cy5.5 was used to label rt-PA, and the prepared EcN-PN/rt-PA-Cy5.5 was administered via tail vein injection. At different time points (0, 0.083, 0.17, 0.33, 0.5, 1, 2, 3, 4, 6, 8, 12, 24, and 48 h), 500 μL of rat orbital venous blood was collected using glass tubes with 0.45 μm aperture and transferred to EDTA anticoagulant tube [52]. The blood samples were centrifuged (3000 rpm, 10 min), and the fluorescence intensity of the supernatant was analyzed using a fluorescence spectrophotometer at excitation and emission wavelengths of 673 nm and 695 nm, respectively. Different pharmacokinetic parameters of two groups were calculated and analyzed by DAS software (version 2.0).

4.11. Construction of thrombus model in vivo

Based on previous research from our group, a in vivo left carotid artery thrombus model was constructed using the ferric chloride induction method [17]. The specific process is as follows: C57BL/6 mice (18 ± 2 g, 8 weeks old) were provided with free access to food and drinking water [55,56]. Under isoflurane anesthesia conditions, the mice's necks were shaved, and the skin was incised with surgical scissors to expose the left common carotid artery after removing surrounding fat and connective tissue. A filter paper (3 × 1 mm) soaked in 10% FeCl₃ solution was placed on the surface of the left common carotid artery bifurcation in mice for 5 min and then removed. After about 10 min, a distinct yellow clot could be observed under a surgical microscope. The wound was sutured, and ultrasound analysis (B Mode, Color Mode, PW Mode, and M Mode) was performed on the mice's necks before and after thrombus formation. After that, the left common carotid artery bifurcation vessels before and after thrombus model construction were removed for sectioning and H&E staining.

4.12. Thrombus targeting analysis of EcN-PN/rt-PA in vivo

Firstly, a mouse left common carotid artery thrombus model was constructed as described above, and the mice were divided into three groups (n = 3). The Cy5.5 probe was used to label rt-PA, and fluorescently labeled rt-PA, PN/rt-PA and EcN-PN/rt-PA (1 × 10⁵ CFU/mL, equal to 0.56 mg/mL rt-PA) for tail vein administration in mice with the dose of rt-PA 2 mg/kg. At different time points (0, 10, 20, 30, 40, 50 and 60 min), the three groups of model mice were placed in a small animal fluorescence imaging device for observation. The fluorescence intensity at the thrombus sites in different groups of mice was recorded at different time points using Image J software. The left carotid artery blood vessels of different groups of mice were harvested, frozen, sectioned, and the different vessel slices were observed under a fluorescence microscope at the 60-min time point.

4.13. Thrombolysis efficacy evaluation of EcN-PN/rt-PA in vivo

A left common carotid artery thrombus model was constructed in C57BL/6 mice (18 ± 2 g, 8 weeks old, n = 5), which were then divided into 3 groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL, corresponding to 0.56 mg/mL rt-PA) with an rt-PA dose of 2 mg/kg. Doppler ultrasound in PW mode was used to analyze the blood flow in the left common carotid artery of mice from different groups before and after thrombus induction and at various times after tail vein administration (control, model, 10, 20, 30, 40, 60, 90, 120, 150, 180, and 210 min). GraphPad software (version 9.0) was used for statistical analysis of blood flow from different groups. After 210 min of treatment, blood vessel tissue from the thrombus site was excised and sectioned from mice in different treatment groups (rt-PA, PN/rt-PA, EcN-PN/rt-PA). Tissue sections were stained using H&E staining kit, Masson staining kit, and Cell membrane staining probe DiR (5 μM). The pathological staining results and fluorescence intensity of the different sections were observed via fluorescence microscopy.

4.14. Safety evaluation of EcN-PN/rt-PA in vivo

In order to investigate the biological safety of EcN-PN/rt-PA micro-robot, hemorrhagic time, blood routine, blood biochemistry, and pathological sections of major organs were assessed in mice treated with EcN-PN/rt-PA micro-robot.

Hemorrhagic time experiment [57]: An in vivo thrombus model was established as previously described. Mice were administered via intravenous injection with different agents (saline, rt-PA, PN/rt-PA, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL, corresponding to 0.56 mg/mL rt-PA) at a dose of 2 mg/kg rt-PA. Immediately after administration, the tails of the mice were cut by 1 cm using surgical scissors, and the cut ends were immersed in physiological saline at 37 °C. The time from hemorrhage onset to cessation was recorded as the hemorrhagic time. The hemorrhagic time for each group was analyzed using GraphPad software (version 9.0).

Blood routine, blood biochemistry and pathology experiments: The in vivo thrombus model was constructed as mentioned above. After 12 h of tail vein administration of different agents (saline, EcN, EcN-PN/rt-PA, 1 × 10⁵ CFU/mL, corresponding to 0.56 mg/mL rt-PA) at a dose of 2 mg/kg rt-PA, blood samples and major organs of mice from different groups were collected. Blood routine and blood biochemistry test kits were used to detect parameter changes in different blood samples. The main organs (heart, liver, spleen, lung, and kidney) of the mice were stained by H&E staining kit and Masson staining kit, and the histological changes in different organs were observed under a microscope.

4.15. Statistical analyze

Statistical comparisons were performed by two-sided unpaired Student's t-test for two-group comparisons. For multiple groups (≥3), one-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test was used to control type I error. Significance levels: P < 0.05, ∗P < 0.01, ∗∗P < 0.001. All data are presented as mean ± SD (n ≥ 3 biological replicates). All statistical analyses were performed by GraphPad Prism software (version 9.0).

Ethics approval

All the animal experiments performed with the approval of the Xiamen University Experimental Animal Center and all procedures performed according to the guidelines and animal welfare protocols. Animal ethics number of Xiamen university animal experiment centre is XMULAC20220249.

CRediT authorship contribution statement

Riyue Liu: Conceptualization, Data curation, Formal analysis, Methodology, Project administration, Software, Writing – original draft, Writing – review & editing. Kexin Zhang: Formal analysis, Methodology, Project administration, Software. Jiale Chen: Methodology, Resources, Software. Nawsherwan: Conceptualization, Investigation, Methodology. Jinrong Zheng: Investigation, Methodology, Project administration, Resources. Yan Wang: Formal analysis, Methodology, Project administration, Software, Supervision, Validation, Writing – original draft. Mangmang Sang: Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Supervision, Visualization, Writing – original draft, Writing – review & editing.

Declaration of competing interest

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 are the Supplementary data to this article:

Data availability

Data will be made available on request.