Abstract

   There is no effective and noninvasive solution for thrombolysis because
   the mechanism by which certain thrombi become tissue plasminogen
   activator (tPA)-resistant remains obscure. Endovascular thrombectomy is
   the last option for these tPA-resistant thrombi, thus a new noninvasive
   strategy is urgently needed. Through an examination of thrombi
   retrieved from stroke patients, we found that neutrophil extracellular
   traps (NETs), ε-(γ-glutamyl) lysine isopeptide bonds and fibrin
   scaffolds jointly comprise the key chain in tPA resistance. A
   theranostic platform is designed to combine sonodynamic and mechanical
   thrombolysis under the guidance of ultrasonic imaging. Breakdown of the
   key chain leads to a recanalization rate of more than 90% in male rat
   tPA-resistant occlusion model. Vascular reconstruction is observed one
   month after recanalization, during which there was no thrombosis
   recurrence. The system also demonstrates noninvasive theranostic
   capabilities in managing pigs’ long thrombi (>8 mm) and in
   revascularizing thrombosis-susceptible tissue-engineered vascular
   grafts, indicating its potential for clinical application. Overall,
   this noninvasive theranostic platform provides a new strategy for
   treating tPA-resistant thrombi.

   Subject terms: Nanoparticles, Nanoparticles, Thrombosis
     __________________________________________________________________

   There is no noninvasive solution for tPAresistant thrombi as the
   mechanism remains obscure. Here, the authors show a constructed
   ultrasound-responsive theranostic platform for real-time monitoring and
   efficient thrombolysis by breaking the keychain in tPA-resistant
   thrombi.

Introduction

   Cardiovascular diseases are among the leading causes of death
   worldwide. Thrombosis can lead to many catastrophic events such as
   pulmonary embolism, myocardial infarction, and stroke^[52]1–[53]3. To
   date, tissue plasminogen activator (tPA) can lead to fibrinolysis by
   transforming plasminogen to plasmin, which made it the most commonly
   used thrombolytic treatment. Besides the short half-life, haemorrahge
   risk and the strict therapeutic time window (generally within 4.5 h) of
   tPA^[54]4–[55]7, tPA can only dissolve ~25% of occluded large
   arteries^[56]5, and despite its fibrinolysis property, 44% of the
   post-treatment thrombi were still fibrin-dominant^[57]8. Consequently,
   fewer than 30% of patients have a satisfactory prognosis^[58]5 due to
   these tPA-resistant thrombi. Although endovascular thrombectomy
   combined with intravenous thrombolysis increased the rate of successful
   recanalization, many patients are ineligible for reperfusion therapies,
   and not all hospitals are endovascular-capable^[59]5,[60]9. Therefore,
   treating these tPA-resistant thrombi is a key clinical challenge and it
   is crucial to investigate a novel noninvasive treatment for
   tPA-resistant thrombi.

   The mechanism underlying the development of tPA resistance in these
   thrombi remains unknown. The components of thrombi themselves may
   result in tPA resistance, for example, fibrin-rich thrombi were
   associated with less sensitivity to tPA^[61]10. During thrombosis,
   under the influence of factor XIII (FXIII), the glutamate and lysine
   residues of fibrin monomers are covalently cross-linked, the resulting
   ε-(γ-glutamyl)lysine isopeptide bonds help consolidate the thrombus to
   withstand the shearing force imposed by blood flow around
   it^[62]11–[63]14, furthermore, this fibrin cross-linking increases the
   resistance of the clot to fibrinolytic systems^[64]15,[65]16.
   Additionally, the altered immune microenvironment enables the
   recruitment of neutrophils that, when activated, produce neutrophil
   extracellular traps (NETs) that prevent the penetration of tPA,
   decreasing treatment efficacy^[66]17,[67]18. Given the important roles
   that NETs and isopeptide bonds may play in the development of tPA
   resistance, it is crucial to thoroughly investigate their presence in
   tPA-resistant thrombi, as a detailed analysis of the composition of
   these thrombi may contribute to the design of more effective
   thrombolytics.

   Recently, thrombolytic strategies based on nanotechnology have been
   proposed^[68]19, including photothermal therapy (PTT)^[69]20,
   photodynamic therapy (PDT)^[70]21 and sonodynamic therapy (SDT)^[71]22.
   Of these, SDT has been considered a promising method due to the deep
   tissue penetration and site-specific features of ultrasound (US). SDT
   refers to the generation of singlet oxygen (^1O[2]) through
   sonosensitizers activated via US, which can achieve thrombolysis
   without the combination of tPA^[72]22,[73]23. Flexible and convenient
   imaging feedback during thrombolysis is important. Currently, the most
   commonly used diagnostic methods are computed tomography, magnetic
   resonance imaging and digital subtraction angiography
   (DSA)^[74]24,[75]25. Although these imaging techniques serve as “gold
   standard” for diagnosis, their use is challenging and costly in terms
   of obtaining timely feedback during treatment. US imaging is a commonly
   used and convenient method in clinical practice. Moreover, its
   resolution can be easily enhanced via the application of functional
   agents^[76]26.

   Here, analysis of clinical thrombi obtained from stroke patients
   revealed that NETs, dipeptides and fibrin comprise the keychain
   responsible for the formation of tPA-resistant thrombi. A
   multifunctional theranostic platform integrating SDT and mechanical
   thrombolysis with real-time US imaging was then designed. This system
   involves activation of the sonosensitizer Rose Bengal (RB) to produce
   abundant ^1O[2], which destroys NETs and isopeptide bonds, while the
   theranostics of thrombolysis were aided by mechanical blasting with
   perfluorohexane (PFH)^[77]27. The constructed platform showed great
   recanalization potential in various occlusion models by destroying the
   keychain in tPA-resistant thrombi, and the recanalized arteries
   achieved functional reconstruction. Taken together, our work provides a
   noninvasive, efficient, and convenient integrated theranostic platform
   as a promising candidate for treating tPA-resistant thrombi.

Results

Construction and characterization of a US-responsive theranostic platform

   An integrated platform for thrombus theranostics based in the US was
   designed. As shown in Fig. [78]2a, hollow mesoporous silica (hmSi)
   submicron particles were synthesized^[79]28.
   CREKA(Cys-Arg-Glu-Lys-Ala), a fibrin-targeting peptide^[80]29,
   covalently binds the hmSi submicron particles via amide bonds, helping
   the particles accumulate at the thrombus. Then, the sonosensitizer RB
   and phase transition molecular PFH are sequentially loaded into of the
   hmSi particle, producing hmSi-CREKA-RB-PFH particles. When stimulated
   by US, RB reacts with O[2] to form ^1O[2]^[81]21, and PFH transitions
   from the liquid to the gas phase^[82]28 (Fig. [83]1).

Fig. 2. Characteristics of hmSi-CREKA-RB-PFH.

   [84]Fig. 2
   [85]Open in a new tab

   a Synthesis scheme for hmSi-CREKA-RB-PFH. hmSi hollow mesoporous
   silica, CREKA Cys-Arg-Glu-Lys-Ala peptide, RB Rose Bengal, PFH
   perfluorohexane. b TEM image of hmSi-CREKA-RB-PFH. TEM transmission
   electron microscope. c N[2] adsorption-desorption isotherms of hmSiO[2]
   (insets: corresponding pore size distribution). d Element mapping for
   Si, N, I, and F. e UV‒VIS absorption spectra and digital images of
   different submicron particles. a.u. refers to absorbance unit. ^1O[2]
   was detected based on f DPBF degradation rate and g changes in SOSG
   fluorescence intensity. n = 3 independent experiments. Data are
   presented as mean ± SEM. Source data underlying graph c, e–g are
   provided as a Source Data file. Each experiment was repeated three
   times or more independently with similar results.

Fig. 1. Schematic illustration of an ultrasound-responsive theranostic
platform.

   [86]Fig. 1
   [87]Open in a new tab

   hmSi hollow mesoporous silica, CREKA Cys-Arg-Glu-Lys-Ala peptide, RB
   Rose Bengal, PFH perfluorohexane; US ultrasound. After injection,
   hmSi-CREKA-RB-PFH particles can accumulate at the thrombus. Under US
   irradiation, RB reacts with O[2] to form ^1O[2], and PFH transitions
   from liquid to gas, thus achieving noninvasive real-time ultrasonic
   imaging and effective thrombolysis.

   Transmission electron microscope (TEM) images and distribution curves
   showed that hmSi-CREKA-RB-PFH was successfully constructed with a
   uniform size of ~200 nm and a mesoporous shell thickness of 50 nm
   (Fig. [88]2b, [89]S1a, [90]b). The obvious hysteresis loop in the N[2]
   adsorption-desorption isotherms of hmSi indicates the typical
   mesoporous structure of the particle (Fig. [91]2c). The surface area
   and pore volume are 809.86 m^2/g and 0.74 cm^3/g, respectively. The
   average hole diameter (insets in Fig. [92]2c) is 5.72 nm, supporting
   the high loading potential of RB and PFH. Further, the elemental
   mappings of Si, N, I, and F indicate the presence of hmSi, CREKA, RB
   and PFH in the submicron particles (Fig. [93]2d). The hydrodynamic size
   of the particle increased from 218 to 230, 271, and 247 nm after the
   binding of CREKA and the incorporation of RB and PFH, respectively. A
   slight decrease may result from the hydrophobicity of PFH
   (Fig. [94]S1c). The corresponding ζ potential changed from 28.9 to
   24.3, −2.3, and 30.8 mV, respectively; the electronegativity of CREKA
   and RB led to a charge transition to negative (Fig. [95]S1d). The
   changes in hydrodynamic size and ζ potential indicated that although
   the vast majority of RB and PFH were loaded inside the carrier, the
   possibility of surface adhesion cannot be ruled out. Additionally, as
   shown in the UV–vis absorption spectra (Fig. [96]2e), there were clear
   and obvious absorption peaks located at 450 nm and 555 nm, which were
   attributed to CREKA (FITC-labeled) and RB, respectively.

   Having assessed the structure of the hmSi-CREKA-RB-PFH particles, we
   next sought to explore their sonodynamic characteristics by verifying
   the generation of ^1O[2] detected with 1,3-diphenylisobenzofuran (DPBF)
   (Fig. [97]2f and [98]S2a) and Singlet Oxygen Sensor Green (SOSG)
   (Fig. [99]2g and [100]S2b). Changes in the signals of these ^1O[2]
   probes were observed only in the presence of RB and the delivery of US,
   demonstrating the superior ^1O[2] production performance of the system
   and indicating its enormous potential in sonodynamic thrombolysis. The
   US-mediated phase transition of PFH was also explored. Digital photos
   confirmed the US-responsive behavior of PFH-encapsulating submicron
   particles, as PFH bubbles were gradually generated and could be
   visualized after 20 min of US irradiation (Fig. [101]S3a and
   Supplementary Movie [102]1). PFH can act as an echo contrast agent,
   increasing the contrast of the US images with increasing US duration
   (Fig. [103]S3b, [104]c).

   Subsequent experiments were conducted to confirm the safety of
   hmSi-CREKA-RB-PFH. According to CCK-8 assays^[105]30, cell viability
   was greater than 90% after incubation for 72 h (Fig. [106]S4a).
   Hemocompatibility is another critical parameter for evaluating the
   suitability of a system for in vivo biomedical applications. Even at a
   high hmSi-CREKA-RB-PFH concentration of 400 µg/mL, negligible hemolysis
   rates (6.54%) were detected (Fig. [107]S4b). We then examined blood
   inflammation after injection of the submicron particles, the proportion
   of lymphocytes increased by 25.1% after thrombosis and remained
   unchanged after continuous injection of hmSi-CREKA-RB-PFH for three
   days (Fig. [108]S4c), indicating that early inflammation may be
   attributed to thrombosis progression and that the submicron particles
   do not further aggravate the inflammatory response. Organ fluorescence
   imaging is used to show the biological distribution of submicron
   particles in major organs (Fig. [109]S4d, [110]e). The fluorescence of
   the injection submicron particles at the blocked vessels demonstrated
   its ability to specifically target thrombus. Additionally, after 3 days
   of injection, the fluorescence signals in the liver and kidney
   suggested that the particles might be excreted by these two organs. On
   day seven following delivery, these signals recovered to the same level
   as the control group. H&E staining demonstrated that it will not cause
   morphological changes in major organs, indicating that it is
   biologically safe (Fig. [111]S4f). These results demonstrated that the
   hmSi-CREKA-RB-PFH submicron particles exhibited excellent
   biocompatibility, supporting their safe systemic administration as part
   of our proposed thrombolytic system.

The US-responsive theranostic platform destroys NETs in thrombi

   During thrombosis, neutrophils respond to the immune changes and
   release web-like DNA structures called NETs. NETs are constitutively
   presented in the thrombi of acute ischemia stroke (AIS)
   patients^[112]17,[113]18. We obtained clinical thrombi from AIS
   patients who have failed intravenous thrombolysis within the
   therapeutic time window (4.5 h), and underwent endovascular
   thrombectomy within 24 hours from symptom onset. The detailed clinical
   information is shown in Supplementary Data [114]1. To rule out the
   effect of heterogeneity among clinical thrombus samples, the thrombi
   were evenly divided into four groups according to the treatments to
   which they were subjected: control, US, US + hmSi-CREKA-RB(SDT), and
   US+ hmSi-CREKA-RB-PFH (combination therapy of SDT and
   mechanical-blasting therapy). First, we found that the US alone
   substantially decreased the concentration of NETs (Fig. [115]S5a).
   Given that photothermal therapy has been reported as a new thrombolysis
   method^[116]20 and considering the potential thermal effects generated
   by continuous US irradiation, we applied an extracorporeal circulation
   device (Fig. [117]S5b) to exclude the influence of heat on NETs
   dissolution. After 30 min of US irradiation, immunofluorescence
   staining revealed significant destruction of NET structures under the
   combination therapy, while lots of NETs-corresponding areas existed in
   the control group, the number of NETs per unit area decreased
   significantly to 10.2% of the control group. (Figs. [118]3a and
   [119]2b).

Fig. 3. The theranostic platform causes lytic NETs.

   [120]Fig. 3
   [121]Open in a new tab

   a Immunostaining of NETs in clinical thrombi following different
   treatments. NETs neutrophil extracellular traps. b NETs/mm^2 in
   clinical thrombi. n = 4 samples. One-way ANOVA with LSD post hoc
   analysis was employed to compare the differences between groups. c SEM
   images of NETs in different treatment groups and the corresponding
   Sytox Green-stained samples visualized at the cellular level. SEM
   scanning electron microscope. Arrows indicate NETs (neutrophils and the
   filamentous structure around the nucleus). d Percentages of NET-like
   structures at the cellular level, n = 3 independent experiments.
   Kruskal–Wallis test was employed to compare the differences between
   groups. e Immunostaining of NETs in rat carotid artery thrombi
   following different treatments. Arrows indicate NETs (DNA colocalized
   with Myeloperoxidase and Cit[3]H[4]). f NETs per unit area in rat
   carotid artery thrombi. n = 7 samples. g Scheme of NETs formation and
   lysis within thrombi. Welch’s ANOVA with Tamhane’s T2 post hoc analysis
   was employed to compare the differences between groups. Data are
   presented as mean ± SEM. Source data underlying graph b, d and f are
   provided as a Source Data file. Each experiment was repeated three
   times or more independently with similar results.

   Next, we sought to examine the effect of US+hmSi-CREKA-RB-PFH at the
   cellular level. Neutrophils were extracted from rat peripheral blood
   and phorbol 12-myristate 13-acetate (PMA) was used to induce NET
   formation. The filamentous structure around nucleus indicated
   successful in vitro induction (Fig. [122]S6). Sytox Green was used to
   label extracellular DNA^[123]31, after 10 min of US irradiation, SEM
   and immunofluorescence staining (Fig. [124]3c) jointly revealed that
   the combination therapy efficiently decreased the proportion of NETs
   from the initial 10.16% to 0.29% (Fig. [125]3d). Considering that the
   mechanical vibration of US may cause the detachment of NETs, we
   analyzed the post-treatment culture medium and discovered a significant
   amount of shed cells in the combination therapy group (Fig. [126]S7a).
   Additional immunofluorescence labeling revealed that the shed cells
   were still attached to undamaged NETs in the US group. The dispersed
   state of the shed cells in the SDT group and the combination therapy
   group suggested that NETs degradation aggravated cell detachment since
   no complete NETs were observed (Fig. [127]S7b). We investigated the
   dose-effect relationship between the generation of ^1O[2] and the
   destruction of NETs and found that the amount of destroyed NETs was
   positively correlated with ^1O[2] (Fig. [128]S8a–[129]c), indicating
   that SDT further mediates NETs damage.

   The efficacy of our system in destroying NETs was ultimately verified
   in a rat carotid artery occlusion model. It was reported that NETs
   peaks at 3–5 days after stroke^[130]32, accordingly, 4 days after
   thrombosis, we identified a large amount of NETs in the model
   (Fig. [131]S9), so we chose the fourth day as the starting time for
   treatment. Immunofluorescence staining showed that the positive signals
   of trichrome colocalization were significantly changed. Lytic traps
   were found in the SDT group, and the fluorescence of NETs completely
   disappeared in the combination group (Fig. [132]3e). Statistical
   results showed that the number of NETs declined from 290 to 13
   per mm^2, indicating high NETs-destruction capacity of the combination
   therapy in vivo (Fig. [133]3f).

   DNaseI can disrupt the DNA skeleton^[134]18 and was set as positive
   control group, the combination therapy group exhibited comparable
   damage to NETs both in vitro (Fig. [135]S10a, [136]b) and in vivo
   (Fig. [137]S10c, [138]d). However, DNaseI can only increase the
   reperfusion rate from 33.73% to 41.58% (Fig. [139]S10e, [140]f),
   indicating the invalid individual destruction of NETs. The above
   results illustrated that the combination therapy can destroy the NETs
   in thrombus (Fig. [141]3g), providing another way to degrade NETs in
   addition to DNaseI, whose efficiency is limited due to its short
   half-life and instability^[142]27.

The US-responsive theranostic platform destroys isopeptide bonds in thrombi

   The isopeptide bond is a key contributor to the stability of fibrin.
   Immunofluorescence analysis of clinical thrombi showed that there
   existed a large number of isopeptide bonds within the tPA-resistant
   thrombi (Fig. [143]4a). After 30 min of treatment, the fluorescence
   area of isopeptide bonds decreased from 9.7% to 0.35% in the
   combination therapy group (Fig. [144]4b).

Fig. 4. The theranostic platform breaks isopeptide bonds.

   [145]Fig. 4
   [146]Open in a new tab

   a Immunostaining of isopeptide bonds in clinical thrombi following
   different treatments. b Percentages of isopeptide bonds in clinical
   thrombi. n = 5 samples. c SEM images and the corresponding
   immunostaining of isopeptide bonds induced in vitro. Grayscale analysis
   of immunofluorescence corresponding to d fibrin and e dipeptides in
   different treatment. f Proportion of isopeptide bonds induced in vitro.
   n = 5 independent experiments. g Immunostaining of isopeptide bonds and
   fibrin in rat carotid artery thrombi following different treatments. h
   Proportions of isopeptide bonds in rat carotid artery thrombi. n = 7
   samples. i Schematic of the formation and breakdown of isopeptide bonds
   in the thrombus. Data are presented as mean ± SEM. Welch’s ANOVA with
   Tamhane’s T2 post hoc analysis was employed to compare the differences
   between groups. Source data underlying graph b, f and h are provided as
   a Source Data file. Each experiment was repeated three times or more
   independently with similar results.

   Then the crosslinks were induced in vitro by treating fibrinogen with
   thrombin and factor XIII (FXIII). FXIII plays an important role in
   regulating the structure of fibrin^[147]14, compared with that in the
   FXIII-deficient group, the diameter of fibrin became larger in the
   FXIII-contained group, which may be attributed to transverse fibrin
   aggregation (Fig. [148]4c). The fibrin network vanished (Fig. [149]4d)
   as the isopeptide bonds weakened (Fig. [150]4e), implying the
   importance of isopeptide bonds in the formation of the fibrin network.
   The results indicated that the US can preliminarily destroy the
   dipeptides, perhaps via structural loosening due to the physical
   vibrations generated by the US waves. By creating ^1O[2], SDT can
   further degrade the dipeptides. Compared with those in the control
   group, the proportion of isopeptide bonds decreased from 12.22% to
   0.02% after combination therapy (Fig. [151]4f). The dose-effect
   relationship between ^1O[2] and the number of dipeptides was similar to
   that between ^1O[2] and the amount of NETs (Fig. [152]S8d),
   demonstrating that the degree of isopeptide bond breakdown is
   associated with the amount of singlet oxygen.

   In the rat carotid thrombosis model, after three consecutive days of
   medication and treatment, the fluorescence area of the isopeptide bond
   was decreased (Fig. [153]4g), and the number of isopeptide bonds in
   four groups was 2.15%, 1.98%, 0.33%, and 0.13%, respectively
   (Fig. [154]4h), indicating an effective destruction of dipeptides in
   vivo (Fig. [155]4i).

   As a key contributor to thrombus insolubility, isopeptide bonds help
   thrombi resist shear stress, and their disruption may result in a more
   soluble thrombus. We extracted the blood vessels that were still
   partially blocked after three applications of treatments and then
   performed a scouring experiment to verify whether the thrombi could be
   dispersed by shear stress (Fig. [156]S11a). The color of the vessel
   lumen became lighter after flushing (Fig. [157]S11b), suggesting that
   the breakdown of isopeptide bonds reduces the stability of the thrombi.
   Additionally, the scoured thrombus fragments were <10 μm
   (Fig. [158]S11c), with no risk of any secondary blockage in distal
   small vessels^[159]21.

The US-responsive theranostic platform promotes degradation of fibrin
scaffold in thrombus

   Fibrin is a major supporting component of the thrombus. To demonstrate
   that the degradation of NETs and isopeptide bonds contributes to
   fibrinolysis, complexes containing fibrin and NETs were induced in
   vitro. NETs in the thrombus prevented the tPA-mediated degradation of
   fibrin, while the introduction of DNaseI promoted fibrinolysis
   (Fig. [160]S12a, [161]b), suggesting that NETs-destruction contributes
   to fibrin degradation. Similarly, the A2LD1 enzyme specifically breaks
   isopeptide bonds^[162]33, loosens the structure of fibrin and enlarges
   the pores when used in isolation. The combination of A2LD1 and tPA
   significantly improved fibrin degradation, demonstrating the
   effectiveness of disrupting the isopeptide bond on fibrinolysis
   (Fig. [163]S12c, [164]d) and, ultimately, on thrombolysis.

   To observe the overall thrombolytic effect, the clinical thrombi were
   stained with haematoxylin and eosin (H&E) (Fig. [165]5a). Subsequent
   histological analysis showed that the porosity in the US +
   hmSi-CREKA-RB-PFH group was twice that in the control group, which
   makes thrombi more conducive to dissolution (Fig. [166]5b). To more
   specifically highlight the fibrin changes following the different
   treatments, we conducted MSB staining, which yielded results consistent
   with those of H&E staining (Fig. [167]S13a), the residual fibrin
   decreased by 28.55% in the combination group compared to the control
   group (Fig. [168]S13b).

Fig. 5. Theranostic platform caused the breakdown of the fibrin scaffold.

   [169]Fig. 5
   [170]Open in a new tab

   a H&E staining of clinical thrombi following different treatments. b
   Measurements of thrombus porosity in different groups. n = 5 samples. c
   Digital images of artificial thrombi subjected to different treatments
   for different durations. d UV absorption peaks of the supernatant at
   405 nm for different durations and for different treatments. n = 3
   independent experiments. a.u. refers to the absorbance unit. e Particle
   size distribution of the supernatant after treatment with the
   artificial thrombus. f Confocal images of fibrin scaffolds at different
   time points in the different treatment groups. g Statistics of residual
   fibrin. n = 5 independent experiments. Data are presented as
   mean ± SEM. one-way ANOVA with the LSD post hoc test was employed to
   compare the differences between groups. Source data underlying graph
   b–e and g are provided as a Source Data file. Each experiment was
   repeated three times or more independently with similar results.

   Subsequently, artificial blood clots were induced for in vitro
   thrombolytic examinations. Visual analysis showed that the volumes of
   blood clots gradually decreased over time and for treatments with
   additional thrombolytic components (Fig. [171]5c). The UV absorption
   peak at 405 nm, attributed to dissolved fibrin^[172]34, served as
   additional evidence (Fig. [173]5d and [174]S14). The particle size
   distribution of the supernatant was further measured to rule out the
   likelihood of small thrombolytic fragments leading to secondary
   embolism, and the fragments of each group were concentrated in
   100–1000 nm. The fragments in the combination group were lager than
   those in the other groups, possibly due to the local mechanical effect
   of PFH blasting, nevertheless, the fragments were less than 3000 nm and
   thus unlikely to potentially block the distal vasculature
   (Fig. [175]5e).

   To more directly characterize the destruction of fibrin, we induced a
   fibrin scaffold in vitro and fluorescently labeled it with CREKA-FITC.
   Confocal imaging showed that the cavity size of fibrin scaffold
   gradually increased (Fig. [176]5f), in line with the results of the
   clinical thrombi. When the US lasted for 10 min, the fibrin in the
   combination therapy group could be successfully dissolved into
   fragments (Fig. [177]5g), and the collapse of the main component is
   important for thrombolysis.

The US-responsive theranostic platform performs well in treating rat thrombi
in vivo

   To evaluate the in vivo efficacy of the combination therapy, a
   Sprague–Dawley (SD) rat carotid artery thrombus model was constructed
   using the ferric chloride (FeCl[3]) chemical injury method^[178]21, and
   treatment was started on the fourth day after thrombosis
   (Fig. [179]6a). Given that clinical thrombi obtained via endovascular
   thrombectomy have undergone intravenous thrombolysis, these thrombi can
   be assumed as tPA-resistant. tPA was first applied on the
   FeCl[3]-induced thrombus model to assess the level of tPA resistance.
   Doppler ultrasonography revealed that blood vessel patency increased
   from 33.73% to only 36.55% (Fig. [180]S15a, [181]b), while H&E staining
   revealed numerous thrombus blockages (Fig. [182]S15c, [183]d),
   indicating the establishment of tPA resistance.

Fig. 6. The therapeutic effect of the US-responsive theranostic platform in
rats.

   [184]Fig. 6
   [185]Open in a new tab

   a Schematic illustration of the experimental timeline. b Real-time
   Doppler ultrasound images in different treatment groups. c
   Laser-speckle imaging of carotid arteries. d Percentage of blood flow
   velocity between blood vessels in different treatment groups and normal
   contralateral blood vessels. n = 8 samples. e Mean blood perfusion of
   carotid artery in different treatment groups. n = 8 samples. f DSA of
   the carotid artery of rats after treatments. The red arrows indicate
   the site of recanalization. g Digital photos of the carotid artery
   after each treatment. h Histopathology of treated vessels and i the
   patency rate of the vessels. n = 8 samples. Data are presented as
   mean ± SEM. one-way ANOVA with the LSD post hoc test was employed to
   compare the differences between groups. Source data underlying graph d,
   e and i are provided as a Source Data file. Each experiment was
   repeated three times or more independently with similar results.

   In vivo experiments with the US+hmSi-CREKA-PFH–based treatment were
   conducted to investigate the mechanical-blasting thrombolytic effect.
   Immunofluorescence revealed obvious NETs (Fig. [186]S16a, [187]b) and
   isopeptide bonds (Fig. [188]S16c, [189]d), and the vascular patency
   rate was 39.42% (Fig. [190]S16e, [191]f). The therapeutic effect of
   mechanical-blasting alone was not ideal, possibly because blasting is
   intended mainly to affect the clot surface. However, PFH play an
   important role on enhancing the contrast of ultrasound imaging in vivo
   (Fig. [192]S17), indicating the necessity of PFH combination.

   Next, we sought to assess the ability of the hmSi-CREKA-RB-PFH particle
   to target the thrombi. The FTC-labeled CREKA emitted an obvious
   fluorescent signal at the site of the thrombus after tail vein
   injection (Fig. [193]S18a). There was no significant difference in
   fluorescence area after 1 h, so all the treatments started 1 h after
   injection (Fig. [194]S18b). After three consecutive days of noninvasive
   treatment (1 MHz, 1.5 W/cm^2, 10 min each time), the patency of blood
   vessels was examined in vivo. Doppler ultrasonography showed obvious
   signal of blood flow in the combination therapy group (Fig. [195]6b),
   which was later confirmed by laser-speckle imaging (Fig. [196]6c). As
   blood flow velocity is correlated with lumen diameter, we calculated
   the ratio of blood flow velocity between the occlusive side and the
   normal control side in the same rat, which gradually increased from the
   initial 11.4% to 88.8% in the combination therapy group (Fig. [197]6d)
   and the mean blood perfusion intensity of carotid artery increased from
   189.9 PU/mm^2 to 498.9 PU/mm^2 (Fig. [198]6e). DSA further supported
   the considerable impact of combination therapy on recanalization
   (Fig. [199]6f and Supplementary Movie [200]2). The same therapeutic
   tendency was observed from the digital images under a microscope
   (Fig. [201]6g). In the combination therapy group, no obvious thrombus
   structure was observed, and the shape and color of the blood vessels
   were close to normal levels. Finally, the vessels were sliced into
   cross-sections, and we calculated the area of thrombus occupying the
   entire lumen, the obtained histological outcomes matched those of
   previous characterizations (Fig. [202]6h), and the patency rate of
   blood vessels in the combination therapy group reached 94.3%
   (Fig. [203]6i), in line with the US results.

   Because shear stress may improve the thrombolytic effect in an arterial
   occlusion model, we further verified the therapeutic efficacy in the
   inferior vena cava (IVC) model to examine the reperfusion effect in the
   absence of shear stress. Immunofluorescence staining revealed that NETs
   were also present in FeCl[3]-induced thrombi in the IVC
   (Fig. [204]S19a). After three applications of the noninvasive
   treatment, Doppler ultrasound imaging showed the recovery of blood flow
   (Fig. [205]S19b). Histological staining showed that the patency rate
   increased from 30.22% to 85.34% (Fig. [206]S19c, [207]d), indicating
   that the combination therapy can achieve satisfactory thrombolysis
   efficiency despite the lack of shear stress in the IVC.

   Additionally, we delayed treatment for 30 days after thrombus formation
   to establish an old thrombus model, as the treatment of old thrombi
   remains a clinical challenge. The CREKA-FITC fluorescence signal after
   injection indicated that hmSi-CREKA-RB-PFH was able to target old
   thrombi (Fig. [208]S20a). Doppler ultrasonography revealed that the
   relative patency rate in the combination therapy group was twice that
   in the control group (Fig. [209]S20b, [210]c). Histological staining
   showed that vessels in the treated group exhibited better blood flow
   patency than those in the untreated groups, ranging from 24.38% to
   63.12% (Fig. [211]S20d, [212]e), indicating the potential of the
   combination therapy for treating old thrombi.

The US-responsive theranostic platform can monitor and maintain the long-term
patency of TEVGs through noninvasive thrombolysis

   Tissue-engineered vascular grafts (TEVGs) can be used for coronary
   artery bypass grafting. However, despite the high clinical demand for
   small-diameter TEVGs, such vascular grafts easily form thrombi,
   ultimately leading to transplantation failure. It is important to
   maintain the long-term patency of small-diameter vessels after
   transplantation^[213]35,[214]36. Given the demonstrated ability of the
   proposed theranostic platform to recanalize a rat occlusion model, we
   applied this thrombolytic system to the long-term maintenance of TEVGs
   (Fig. [215]7a). Engineered acellular blood vessels were transplanted
   into the carotid arteries of rats. Blockage of TEVGs after
   transplantation was preliminarily determined by Doppler
   ultrasonography, weak blood flow signals indicate a blockage in the
   lumen (Fig. [216]7b). After three applications of treatment, blood
   perfusion was effectively restored, and the combination therapy
   achieved 63.06% recanalization (Fig. [217]7c and Supplementary
   Movie [218]3). The digital photos were shown in Fig. [219]7d, there was
   a macroscopic thrombotic clot in control group, while the vessel
   morphology following application of the combination therapy resembled
   that initially observed after transplantation. The H&E staining further
   verified that the vascular patency rate was 76.45% (Fig. [220]7e,
   [221]f). Taken together, these results show that our proposed
   US-responsive theranostic platform demonstrates good performance in the
   treatment of thrombi in small-diameter TEVGs. This technique could thus
   serve as a noninvasive approach to vascular graft thrombolysis, which
   is critical for maintaining the long-term patency of small-diameter
   vessels and facilitating their clinical use.

Fig. 7. US-responsive theranostic platform maintains patency in TEVGs.

   [222]Fig. 7
   [223]Open in a new tab

   a Schematic illustration of the blockage and recanalization following
   TEVG transplantation in the control and combination therapy groups.
   TEVG, tissue-engineered vascular graft. b Real-time Doppler
   ultrasonography, and c the percentage of blood flow velocity between
   TEVGs and normal contralateral blood vessels before and after
   treatment. n = 5 samples. Two-tailed paired samples t test was employed
   to compare the differences between groups. d Digital photos of normal,
   blocked, and recanalized TEVGs. e Histopathological staining and f the
   corresponding patency rates of the TEVGs. n = 5 samples. Two-tailed
   independent samples t test was employed to compare the differences
   between groups. Data are presented as mean ± SEM. Source data
   underlying graphs c and f are provided as a Source Data file. Each
   experiment was repeated three times or more independently with similar
   results.

Long-term monitoring of functional reconstruction of recanalized vessels

   To further elucidate the long-term therapeutic efficacy of our
   combination therapy, blood samples and the recanalized vessels one day
   and one week after treatment were harvested. Hematological analysis
   showed that the injection of hmSi-CREKA-RB-PFH and the subsequent US
   treatment process did not further exacerbate the inflammatory response
   caused by thrombosis (Fig. [224]S4d and [225]S21a). Although platelets
   were slightly activated (2.11%) on the first day after treatment, this
   phenomenon subsided by the seventh day after recanalization
   (Fig. [226]S21b), during which no secondary embolism was observed
   (Fig. [227]S21c). Histological staining of major organs revealed no
   long-term toxicity in the combination therapy group (Fig. [228]S21d).
   RNA sequencing analysis of the recanalized vessels revealed that over
   time, two hundred thirty-six genes were upregulated, whereas two
   hundred sixteen genes were downregulated with respect to the control
   group. A volcano plot was generated to visualize the differentially
   expressed genes (Fig. [229]8a) and revealed robust downregulation of
   the inflammation-related genes Pi15, Ptgs2 and Tfpi2, which is
   consistent with the results of Kyoto Encyclopedia of Genes and Genomes
   (KEGG) pathway enrichment analysis (Fig. [230]S22a).
   Thrombosis-mediated damage increased the expression of IL6 and CCL2,
   which recruited macrophages and transferred the blood vessels into an
   acute inflammation stage. Notably, the expression of the
   anti-inflammatory cytokine IL-10, and homeostasis-related genes
   Serpine1, Tmip1, Ptx3, and Tsp1^[231]37 all exhibited phased increases
   (Fig. [232]S22b), indicating the timely response of the compensatory
   mechanism. In addition, gene ontology (GO) analysis revealed that a
   great number of top 30 difference terms were related to tissue
   development (Fig. [233]8b and [234]S22c), specifically,
   development-related genes, such as Bmp6, Sox10, Myh1 and Angpt1, were
   upregulated (Fig. [235]8a), reflecting the remodeling of the
   recanalized vessels. Gene set enrichment analysis (GSEA) also revealed
   that genes related to the development of vascular smooth muscle were
   upregulated (Fig. [236]S22d), immunofluorescence staining revealed that
   the reconstruction of smooth muscle in recanalized vessels approached
   that in normal blood vessels 28 days later (Fig. [237]8c, [238]d).

Fig. 8. Reconstruction of recanalized vessels treated with US-responsive
theranostic platform.

   [239]Fig. 8
   [240]Open in a new tab

   a Volcano plots showing the identified upregulated and downregulated
   genes one day and one week after combination therapy
   (|fold-change| ≥ 1.0 with q value < 0.05). n = 3 samples. b GO
   biological process analysis of genes significantly changed between the
   two groups (|fold-change| ≥ 1.0 with P value < 0.05) using Fisher’s
   exact test (two-sided). The q values shown represent the adjusted
   p-values for multiple comparisons, corrected using the
   Benjamini-Hochberg procedure. GO, gene ontology. c Immunofluorescence
   staining of CD31 and α-SMA. The number of cells expressing d α-SMA, e
   CD31, f ZO-1, and g eNOS. n = 8 samples. Data are presented as
   mean ± SEM. Kruskal–Wallis test was employed to compare the differences
   between groups. Source data underlying graph d–g are provided as a
   Source Data file. Each experiment was repeated three times or more
   independently with similar results.

   Given the importance of endothelial cells (ECs) in maintaining
   long-term vessel patency, the morphology of ECs and their biological
   functions were tested. ZO-1, a tight-junction protein, indirectly
   reflects EC structure. No cells expressing α-SMA and no ECs were
   observed in the control group (Fig. [241]S23a–[242]f). In the
   recanalized vessels, almost no ECs were observed at the completion of
   treatment due to the damage caused by thrombosis (Fig. [243]8c,
   [244]e). ECs with poor continuity and large gaps were observed after 7
   days of reperfusion. Finally, a distinct layer of ECs with tight
   connections was visible at 28 days (Fig. [245]8f and [246]S23g). The
   number of ECs per unit area increased with time (Fig. [247]8e),
   indicating that unobstructed blood flow is important for vascular
   reconstruction. As an important enzyme in cardiovascular system,
   endothelial nitric oxide synthase (eNOS) can catalyze the generation of
   nitric oxide (NO) to regulate vasomotion. After treatment, blood
   vessels were harvested at 0, 7, and 28 days for eNOS immunofluorescence
   staining, the results revealed that the number of cells expressing eNOS
   was consistent with the number of ECs (Fig. [248]8g and [249]S23h).

The US-responsive theranostic platform dissolves long thrombi in pig
occlusion model

   We next constructed occlusion models in the pig femoral artery
   (Fig. [250]9a)^[251]31, producing obvious thrombosis (Fig. [252]9b).
   Then, hmSi-CREKA-RB-PFH was injected through the ear vein, after which
   US examination was conducted. Doppler ultrasonography revealed that the
   blood flow signal of healthy blood vessels vanished after thrombus
   formation (Fig. [253]9c), and the blood flow velocity increased from
   45.85 cm/s to 99.53 cm/s as the blood vessel became narrower. After
   three, 10-min US sessions, the blood flow signal recovered, and the
   flow velocity decreased to 46.22 cm/s (Fig. [254]9d and Supplementary
   Movie [255]4), approaching the velocity in normal blood vessels. Blood
   samples from the normal, thrombosis, and thrombolysis periods were also
   collected to measure the level of D-dimer, a specific marker of fibrin
   degradation^[256]32. There was no significant difference in the D-dimer
   level between the normal and thrombosis periods. However, afterwards,
   the D-dimer level increased 20-fold (Fig. [257]9e), indicating that the
   thrombus was effectively dissolved. The results of H&E staining also
   showed the degree of vascular occlusion (Fig. [258]9f), with a
   corresponding increase in the relative patency rate from 17.47% to
   84.69% (Fig. [259]9g). Immunofluorescence staining presented a large
   number of isopeptide bonds in the control group. However, the
   fluorescence of these crosslinks in the treatment group decreased, and
   fibrin was enriched in the position where the isopeptide bond had not
   yet been destroyed (Fig. [260]S24). In addition to the excellent
   recanalization effect, we witnessed contrast improvement induced by the
   transition of PFH in carotid artery clot. The morphology of the
   thrombus became clearer and the resolution of US imaging was enhanced
   over time (Fig. [261]S25), indicating the feasibility of
   hmSi-CREKA-RB-PFH submicron particles as a contrast-enhancing agent for
   real-time monitoring of thrombosis treatment.

Fig. 9. The US-responsive theranostic platform has a therapeutic effect in
pigs.

   [262]Fig. 9
   [263]Open in a new tab

   a Digital photos of an exposed femoral artery. b Real-time US (B-mode)
   and c Doppler ultrasound images showing thrombosis and thrombolysis in
   a femoral artery. d Real-time assessment of blood flow velocity. n = 3
   samples. Two-tailed paired samples t-test was employed to compare the
   differences between groups. e Changes in D-dimer before and after
   treatment. n = 3 samples. Two-tailed paired samples t-test was employed
   to compare the differences between groups. f Histopathological staining
   and g the corresponding patency rate. n = 3 samples. Two-tailed
   independent samples t-test was employed to compare the differences
   between groups. Data are presented as mean ± SEM. Source data
   underlying graph d, e, g are provided as a Source Data file. Each
   experiment was repeated three times or more independently with similar
   results.

Discussion

   The limitations of tPA, a commonly used thrombolytic drug, result in a
   large number of tPA-resistant thrombi, which encourages the development
   of a new tPA-independent noninvasive thrombolysis strategy. Through
   comprehensive literature analysis and our experimental results, we
   found that the NETs–dipeptide–fibrin keychain is primarily responsible
   for the relative insolubility of these thrombi.

   We constructed an integrated theranostics platform for efficiently
   dissolving of tPA-resistant thrombi and providing flexible ultrasound
   imaging feedback. Ultrasound imaging is a flexible and convenient
   method commonly used in clinical practice. In vitro, we found that the
   quantity of NETs and isopeptide bonds was also significantly reduced by
   US alone; we suspect that this efficacy might be related to the thermal
   effect of these sound waves, and we subsequently verified this
   hypothesis via a flowing water system, suggesting that the effect of US
   alone would be limited and that effective thrombolysis would require a
   combination of more powerful methods (such as the proposed platforms,
   which are sensitive to US). Additionally, in rat and pig trials, our
   designed theranostic platform has demonstrated good thrombolytic
   capability, but it may still be challenging in clinical application.
   For example, deep vein thrombus is usually shielded by solid tissue or
   bone, which makes noninvasive ultrasound access difficult. However, the
   depth of tissue penetration of ultrasound depends on experimental
   parameters such as intensity, amplitude, focus, beam uniformity and
   effective radiating of transducer head^[264]38. For instance,
   high-intensity focused ultrasound has been used to treat tumors that
   are deeper than 10 cm^[265]39. We believe that ultrasound-based
   therapies offer a great deal of potential for clinical deep-tissue
   thrombosis.

   Encouragingly, our proposed treatment resulted in a reperfusion rate of
   >90% in a rat carotid thrombosis model. Based on the results of the in
   vivo experiments, we not only identified a new candidate integrating
   noninvasive diagnosis and treatment for thrombolysis but also
   demonstrated its universality across multiple scenarios, including the
   IVC thrombi in rats, the long thrombi in pigs and the spontaneous
   thrombi in TEVGs. At late time points after recanalization, assessment
   of hematological parameters and Bulk RNA analysis revealed that the
   recanalization led to the subside of thrombosis-induced inflammation.
   Although the rebuilding of smooth muscle and ECs suggests a positive
   therapeutic outcome, vascular reconstruction takes time, indicating the
   necessity of long-term anticoagulation after thrombolysis. The use of
   thrombolysis in combination with anticoagulants, such as heparin, may
   be promising in clinical for preserving long-term vascular patency and
   preventing thrombus recurrence. In addition, the risk of secondary
   distal vessel occlusion following endovascular thrombectomy for major
   vessel occlusions has been reported. Although our in vitro experiments
   suggested a low risk of distal small vessel occlusion, due to the
   unpredictability of the site of occlusion following thrombus
   detachment, we didn’t thoroughly examine this risk through histological
   evaluation.

   In conclusion, we analyzed tPA-resistant thrombi retrieved from stroke
   patients to identify a suitable treatment method for these clots. We
   propose that NETs, isopeptide bonds and fibrin jointly constitute the
   keychain leading to tPA resistance. As a noninvasive thrombolytic
   method, the combination therapy can reach the arteries and show great
   thrombolysis efficacy in various occlusion models, while avoiding the
   risk of secondary embolism. Together, these findings highlight the
   favorable clinical application prospects of our US-responsive
   theranostic platform for treating tPA-resistant thrombi.

Methods

Ethical statement

   The study was approved by the Ethics Committee of the First Affiliated
   Hospital of Army Medical University [(A) KY2021023]. Written informed
   consents were obtained from the patients. No compensation was given.
   All animal experiments were approved by the animal ethics committee of
   the Third Military Medical University (approval No. SYXK(Yu) 20170002
   approval time: 2020.4.20).

Materials

   Hexadecyl trimethyl ammonium bromide (CTAB), tetraethylorthosilicate
   (TEOS), 3-aminopropyltriethoxysilane (APTES), PFH and ferric chloride
   (FeCl[3]·6H2O) were obtained from Shanghai Aladdin Biochemical
   Technology Co., LTD. N-hydroxysuccinimide (NHS) was purchased in
   Shanghai Macklin Biochemical Technology Co., Ltd.
   1-ethyl-3(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) was
   obtained from Beijing Solarbio Science & Technology Co., Ltd. DPBF and
   RB were purchased in Shanghai yuanye Bio-Technology Co., Ltd; Green
   SOSG was purchased from Thermo Fisher Scientific. Targeted peptide
   CREKA was custom-made by ChinaPeptides Co. Ltd. All reagents are used
   directly without further processing.

Synthesis of hmSi-CREKA-RB-PFH submicron particles

   The hmSi was synthesized according to the previous literature with some
   modifications^[266]40. Briefly, ethanol (54 mL), water (360 mL) and
   CTAB (900 mg) were mixed evenly overnight, then NaOH (2 mM, 1 mL) was
   added and stirred for 0.5 h, 3 mL TEOS was added dropwise, after
   stirring for 24 h, 60 μL APTES was added and reacted for another 24 h.
   The product was collected by centrifugation (2000 × g, 10 min) and
   washed repeatedly with ethanol and water three times. Then, the sample
   product was dispersed in anhydrous ethanol (60 mL), HCl (120 μL) was
   added, and heated to 60 °C for 6 h, collected by centrifugation and
   water washing (2600 × g, 15 min), and finally dispersed in water.

   FITC-labeled CREKA (20 mg) was dispersed in 5 mL MES buffer (pH = 6.0),
   then EDC (16 mg) and NHS (9 mg) were added to the solution and stirred
   for 30 min to activate carboxyl groups in CREKA. Next, the hmSi (2 mL,
   5 mg/mL), whose surface is rich in amino groups, were mixed with the
   above solution and reacted at 25 °C overnight. The hmSi-CREKA was
   collected by centrifugation (2600 × g, 10 min) and washed three times
   with ethanol. Subsequently, hmSi-CREKA (2 mL, 5 mg/mL) was mixed with
   RB solution (2 mL, 1 mg/mL) overnight and collected by centrifugation
   (2600 × g, 10 min). hmSi-CREKA-RB freeze-dried powder was obtained by
   removing the solvent. 200 mg powder was divided into 600 μL PFH and
   placed in a 5 mL sealed bottle. It was subjected to ultrasound (40 KHz,
   70 W/cm2) in ice water for 3 min, followed by 3 mL of water solution to
   seal the PFH and the final hmSi-CREKA-RB-PFH was stored at
   4 °C^[267]28.

Characterization

   TEM (FEI Talos F200S) was used to characterize the morphology and
   element mapping of hmSi-CREKA-RB-PFH. Dynamic light scattering (DLS)
   was used to measure the size distribution and ζ potential of submicron
   particles with a concentration of 1 mg/mL; The size distribution of
   submicron particles (hmSi, hmSi-CREKA, hmSi-CREKA-RB,
   hmSi-CREKA-RB-PFH), were measured in aqueous solution. The measurement
   of ζ potentials was performed in 1 mM KCl (Zetasizer Nano ZS, Malvern
   Instruments, UK). Test temperature of the instrument was set at 25 °C,
   each sample was tested for five times. Ultraviolet-visible (UV–vis)
   absorption spectra were obtained by a 2600 UV–vis–NIR spectrophotometer
   (Unico, UV-4800). The generation of ^1O[2] was quantified with a
   Microplate Reader (BioTek). The N[2] adsorption-desorption isotherms
   were determined by a fully automatic specific surface and porosity
   analyzer (BeiShiDe, China). In vitro ultrasound processing uses a
   LIPUSTIM® Sonodynamic Therapy System (ShengXiang, SDT-S019).

ROS detection in vitro

   The production of ^1O[2] was determined with a spectrometric method.
   Four different groups were set: Indicator + US, Indicator +
   hmSi-CREKA-PFH + US, Indicator + hmSi-CREKA-RB-PFH, and Indicator +
   hmSi-CREKA-RB-PFH + US. DPBF and SOSG were chosen as two independent
   indicators. Absorption value changes of DPBF reflect the singlet oxygen
   production under different treatment conditions^[268]41. With DMSO as
   the solvent, the initial UV absorption value of DPBF at 417 nm was
   adjusted to ~2.0, the solution was measured every 5 min in a 20 min
   test. Additionally, fluorescence intensity changes of SOSG at was
   measured at 526 nm^[269]42. The time interval and total test time are
   consistent with the UV absorption method.

In vitro cytotoxicity

   The HUVECs were cultured in DMEM medium (96-well plate, 10^5/well) in a
   cell incubator (37 °C, 5% CO[2]). Fresh medium containing
   hmSi-CREKA-RB-PFH at different concentrations (0, 50, 100, 200, 300,
   and 400 µg/mL) was added for 24, 48 and 72 h. Following a standard
   protocol of CCK-8 assay, the absorbance was measured at 450 nm by using
   a microplate spectrophotometer (BioTek).

Hemolysis assay

   Red blood cells (RBCs) were obtained from fresh blood of rats by
   centrifugation (1000 × g, 10 min). Subsequently, hmSi-CREKA-RB-PFH was
   added to the suspension of RBCs at different concentrations (0, 50,
   100, 200, 300, and 400 µg/mL). The mixture was incubated in a cell
   incubator for 1, 2, and 3 h, respectively. The absorption values of the
   collected supernatant were determined at 577 nm with a microplate
   spectrophotometer (BioTek). The hemolysis ratio of RBCs was calculated
   using the following formula: Hemolysis (%) = (OD[sample-background]
   − OD[negative control])/(OD[positive control] − OD[negative
   control]) × 100%. PBS and distilled water were used as the negative and
   positive controls, respectively. The background absorption was the
   absorption of hmSi-CREKA-RB-PFH at different concentrations.

In vivo toxicity assay

   0.2 mL hmSi-CREKA(FITC)-RB-PFH (5 mg/mL) was injected into rats with
   carotid thrombus through tail vein for three days. The major organs of
   rats were harvested on day 0 and day 7 after injections for ex vivo
   fluorescence imaging (Vilber, Newton 7.0 Bio) and H&E staining,
   including heart, liver, spleen, lung and kidney. The vessels with
   thrombus were taken as positive control. Fluorescence intensity were
   analyzed in Kuant analysis software.

   For the four different treatment groups, after three days of continuous
   dosing and treatment, the organs were harvested for H&E staining. Rats
   in the combination therapy group (hmSi-CREKA-RB-PFH + US) were kept for
   one month after treatment for a long-term toxicity study, whose organs
   were collected for H&E staining.

Immunofluorescence staining of clinical thrombi

   Clinical thrombi were obtained from AIS patients who have failed
   intravenous thrombolysis within the therapeutic time window (4.5 h),
   and underwent endovascular thrombectomy within 24 hours from symptom
   onset. Clinical thrombi were evenly divided into four parts before
   conducting the following groups: control, US, US+hmSi-CREKA-RB and
   US+hmSi-CREKA-RB-PFH. The thrombi were incubated in 0.5 mL PBS, the
   latter two groups need to dissolve corresponding submicron particles
   (2 mg/mL). After 30 min US (frequency: 1 MHz ± 5%, pulse period:
   1000 ms, duty ratio: 50%, and the power density: 1.5 W/cm^2), the
   thrombi were immediately fixed with paraformaldehyde and cut into
   20-micrometre-thick frozen sections for immunofluorescence staining.

   The following primary antibodies were used: mouse anti-MPO (1:100,
   Proteintech, 66177-1), rabbit anti-histone H4 (citrulline 3) (1:200,
   Millipore, 3517971), rabbit anti-N epsilon gamma glutamyl Lysine
   antibody (1:200, abcam, ab424), and mouse anti-fibrinogen (1:100,
   abcam, ab34269). After two nights’ incubation at 4 °C, the secondary
   antibodies were used as follows: Alexa Fluor 488 goat anti-mouse IgG
   (1:1000, Invitrogen), Alexa Fluor 568 goat anti-rabbit IgG (1:1000,
   Invitrogen). Staining was visualized with a confocal microscope (ZEISS
   LSM 800).

   Images were processed using specialized software packages (ZEN Viewer
   and ImageJ).

   The number of NETs was determined by manually counting MPO-positive and
   Cit[3]H[4]-positive nuclear in the traced area (0.102 mm^2)^[270]43.

Characterization of NETs in vitro

   Peripheral neutrophils were isolated from the whole blood of rats using
   a separation kit (TBD, LZS1091), and then, they were evenly seeded on
   poly-L lysine-coated coverslips. The neutrophils were treated with PMA
   (Phorbol 12-myristate 13-acetate, Solarbio, P6741, 10 ng/mL). After 4 h
   incubation in a CO[2] incubator, coverslips were taken for four kinds
   of treatments: control, US, US+hmSi-CREKA-RB, and US+hmSi-CREKA-RB-PFH.
   The coverslips were immersed in PBS, the latter two groups need to
   dissolve corresponding submicron particles (2 mg/mL). Ultrasound
   irradiation (1.0 MHz, 1.5 W/cm^2) was performed for 5 min. DNaseI
   (10 µg/mL) treatment served as positive control. Subsequently, cells
   were fixed in 4% paraformaldehyde for 30 min at room temperature and
   then incubated with SytoxGreen (1:10, Beyotime, C1070S) for 1 h at
   37 °C, the nuclei was stained with Hoechst 33342 (1:10,000, Beyotime,
   C1028). Samples were analyzed using a confocal microscope (ZEISS LSM
   800), and coverslip edges were avoided.

Characterization of ε-N-(γ-glutamyl)-lysyl crosslinks in vitro

   200 µL fibrinogen (Sigma Aldrich, F3879, 1 mg/mL) was mixed with 40 µL
   CaCl[2] (Sigma Aldrich, 10043-52-4, 25 mM) and 2.5 μL transglutaminase
   (ProSpec-Tany, ENZ-394-a, 1 mg/mL), 40 µL thrombin (Sigma Aldrich,
   9002-04-4, 250 NIH U/mL) was added quickly. The mixed solution was
   transferred into poly-L-lysine-coated coverslips and incubated in a
   cell incubator for 2 h. Then, the coverslips were taken out and divided
   into four groups for in vitro experiments. After 10 min US, the
   dipeptides were fixed with paraformaldehyde for subsequent
   characterization.

   The coverslips were stained with CREKA-FITC at 37 °C for 2 h. After
   being washed with PBS, the coverslips were incubated with the Anti-N
   epsilon gamma glutamyl Lysine antibody (1:200, abcam, ab424) at 4 °C
   overnight; The secondary antibody was Alexa Fluor 568 goat anti-rabbit
   IgG (1:1000, Invitrogen). Staining was visualized using a confocal
   microscope (Zeiss LSM 800). Images were processed using ZEN Viewer and
   ImageJ.

Characterization of fibrin in vitro

   Fibrin clots were generated on poly-L-lysine-coated coverslips by
   incubating fibrinogen with thrombin and CaCl[2]. Subsequently, fibrin
   coverslips were divided into four groups as before, immersed in PBS
   solution containing different submicron particles (2 mg/mL). Ultrasound
   irradiation (1.0 MHz, 1.5 W/cm^2) was performed for 5 min and 10 min.
   After incubation with CREKA-FITC at 4 °C overnight, the fibrin network
   was observed using a confocal microscope (ZEISS LSM 800).

Dose-effect relationship between ^1O[2] and NETs/dipeptides destruction

   UV absorption spectra of DPBF with different concentrations were
   tested, and the relationship between concentrations and absorbance at
   417 nm was established. Then hmSi-CREKA-RB with a certain RB absorption
   value (A[564] = 1.76) was selected and mixed with DPBF to test the
   absorbance value of DPBF at 417 nm after different ultrasonic time, the
   relative amount of singlet oxygen generated was calculated by the
   absorbance value changes of DPBF (^1O[2]:DPBF = 1:1, only the amount of
   ^1O[2] that reacts with DPBF could be reflected, so it is defined as
   the relative concentration of singlet oxygen).

   NETs and isopeptide bonds were induced in vitro, and a certain
   concentration of hmSi-CREKA-RB (A[564] = 1.76) was added. The content
   of NETs and isopeptide bonds were quantified by ELISA at different time
   points: ① samples of NETs were added into 96-well plates coated with
   mouse anti-MPO antibody (1:100, Proteintech, 66177-1), followed by
   incubation with rabbit anti-histone H4 (citrulline 3) (1:200,
   Millipore, 3517971). ② samples of isopeptide bonds were added into
   96-well plates coated with rabbit anti-N epsilon gamma glutamyl Lysine
   antibody (1:200, abcam, ab424); After labeled with peroxidase (1:1000,
   ZSGB-Bio, ZB-2301), the absorbance at 405 nm was measured by Microplate
   Reader. The relationship between the relative concentration of singlet
   oxygen and the content of NETS or isopeptide bonds was performed on
   Origin Software (ver 2021).

Characterization of NETs-fibrin complex in vitro

   NETs were induced in vitro as above mentioned, 200 μL fibrinogen (Sigma
   Aldrich, F3879, 1 mg/mL) was mixed with 40 μL CaCl[2] (Sigma Aldrich,
   10043-52-4, 25 mM) and 2.5 μL transglutaminase (ProSpec-Tany,
   ENZ-394-a, 1 mg/mL), 40 μL thrombin (Sigma Aldrich, 9002-04-4, 250
   NIH U/mL) was added quickly. After incubating in a 37 °C incubator for
   2 h, the coverslips were taken out and divided into 5 groups for in
   vitro experiments: Control, tPA, DNaseI, DNaseI+ tPA and
   US+hmSi-CREKA-RB-PFH.

   For drug treatments, DNaseI (Roche, 10104159001, 10 µg/mL), tPA (Sigma
   Aldrich, T0831, 1 µg/mL) and DNaseI+ tPA were added respectively and
   incubated at 37 °C incubator for 1 h. For the combination therapy group
   (2 mg/mL hmSi-CREKA-RB-PFH in PBS), 10 min ultrasound irradiation was
   performed.

   The complex was fixed with paraformaldehyde and treated with CREKA-FITC
   at 37 °C for 2 h. After being washed with PBS, the nuclei were stained
   with Hoechst 33342 (1:10,000, Beyotime, C1028). Samples were analyzed
   using a confocal microscope (ZEISS LSM 800), and coverslip edges were
   avoided.

Characterization of artificial thrombus in vitro

   Artificial thrombus was prepared by adding 4 µL of thrombin
   (20 U mL^−1) and 6 µL of CaCl[2] (25 mM) into 30 µL of SD rat blood at
   37 °C for 3 h. The prepared clots were then randomly divided into four
   groups: control, US, US+hmSi-CREKA-RB, and US+hmSi-CREKA-RB-PFH, the
   latter two groups need to dissolve corresponding submicron particles
   (2 mg/mL). The UV–VIS absorption at 504 nm of the supernatant was
   executed, and the blood clot was taken out for photographs every 5 min
   in a 30 min ultrasound irradiation (1.0 MHz, 1.5 W/cm^2), the final
   supernatant was collected for size-distribution analysis (Zetasizer
   Nano ZS, Malvern Instruments, UK).

Animals

   To better control variables, male SPF-grade SD rats aged 6–8 weeks and
   male domestic pigs weighing 60–80 kg were purchased from the
   experimental animal center of the Third Military Medical University
   (license No. scxk (Yu) 2017-0002). All rats were raised in an SPF
   animal room with 12 h light/dark cycle and provided ad libitum access
   to a standard rodent diet (Jiangsu Xietong Pharmaceutical
   Bio-engineering Co., Ltd, 1010088). All pigs were raised in a single
   cage under conditions of 22 °C, 40–50% humidity with a 12 h light/dark
   cycle and had free access to water and food. Pentobarbital sodium
   (100–200 mg/kg) was given for euthanasia in all animals. All
   experimental operations were performed in accordance with the
   guidelines for experimental animals.

Therapeutic efficacy in rat thrombus model

   Male SD rats of ~200 g were injected intraperitoneally with 2%
   pentobarbital sodium (50 mg/kg). After anesthesia, the neck hair was
   shaved. Under sterile conditions, the common carotid artery was
   separated and rinsed with normal saline. Subsequently, the blood vessel
   was covered with filter paper strips (5 mm × 2 mm) that were soaked in
   10% ferric chloride solution for 8 min^[271]21; the blood vessels were
   rinsed with normal saline 5 min later, and intravascular thrombosis was
   observed.

   Four groups were set for in vivo experiments: Control, US,
   US+hmSi-CREKA-RB, and US+hmSi-CREKA-RB-PFH. On the fourth day after
   thrombosis, submicron particles (0.2 mL, 5 mg/mL) were injected through
   a tail vein for 3 days, and ultrasound (1.0 MHz, 1.5 W/cm^2) was
   performed for 10 minutes after 1 hour of each injection. tPA (10 mg/kg)
   or DNAseI (1000 U) were used as controls through tail vein
   injection^[272]44,[273]45.

   Blood flow signals were observed with Doppler ultrasound (VINNO
   6VET/6LAB). The long-axis and short-axis images of the vascular
   thrombus were collected in two-dimensional (2D) and colour flow (CF)
   Doppler modes, respectively. A laser-speckle vascular-function
   evaluation workstation (PeriCam PSI-ZR, Sweden) was used to collect
   blood perfusion images. The patency rate was the percentage of pore
   areas occupying the entire vascular lumen through H&E staining.

TEVG transplantation

   The detailed technical procedure of TEVGs preparation and
   transplantation had been published previously^[274]35. Briefly, carotid
   arteries were harvested from SD rats (250–300 g), decellularized rat
   carotid arteries were obtained by incubating with 0.05% trypsin at
   37 °C for 30 min, and stored at 4 °C for later use. After vascular
   transplantation, the thrombosis was determined by Doppler
   ultrasonography, and hmSi-CREKA-RB-PFH (0.2 mL, 5 mg/mL) was injected
   through tail vein, 10 min US (1.0 MHz, 1.5 W/cm^2) was performed 1 h
   later and repeated for 3 days.

Bulk RNA sequencing

   The recanalized vessels were extracted a day and a week after
   combination therapy. Total RNA was extracted and RNA quality was
   detected. After constructing the general transcriptome library,
   sequencing was performed on the Illumina NovaSeq 6000 (Berry Genomics
   Corporation, Beijing, China). For differentially expressed gene (DEG)
   analysis, significantly changed genes were selected
   (|fold-change | ≥ 1.0 with q value < 0.05) by edgeR v3.3.81. GO and
   KEGG functional enrichment (over representation) of DEGs was analyzed
   using an R package clusterProfiler v4.4.4. GSEA of all genes were also
   conducted using the R package clusterProfiler. Normalized enrichment
   scores were acquired using gene set permutations 1000 times, and a
   cutoff P value of 0.05 was used to filter the significant enrichment
   results.

Platelet activation and blood inflammatory test

   Blood was collected at first and seventh day after treatment to extract
   platelets. EDTA-anticoagulated blood was centrifuged at 500 × g for
   20 min to isolate platelets. Platelet-inhibition buffer was prepared as
   PBS buffer containing 1 mM EDTA, 2 µM prostaglandin E1 (PGE1, Sigma
   Aldrich) and protease inhibitor (Beyotime, P1005). The supernatant was
   collected and mixed with platelet-inhibition buffer at a ratio of 1:1
   to prevent platelet activation. Platelets were then collected by
   centrifugation at 800 × g for 20 min at room temperature and then
   stained with anti-mouse/rat CD62P-PE (1:30, BioLegend, 148305) for
   30 min and washed with 1 ml PBS. Activated platelets were measured on a
   BD LSRFortessa™ flow cytometer. EDTA-anticoagulated blood from control
   group (without treatment) was set as a negative group, and the
   platelets stimulated with 5 μM PMA for 1 min at 37 °C while stirring
   (300 × g) was set as a positive group. During this period, the patency
   of recanalized vessels on the first and seventh day after treatment was
   determined by Doppler ultrasonography.

   For the inflammatory response, the blood at the same time point is sent
   to the Laboratory Department of the Second Affiliated Hospital of Army
   Medical University Hospital for blood routine tests.

Functional verification of the recanalized vessels

   The treated vessels and the contralateral normal blood vessels were
   harvested on Day 0, Day 7 and Day 28 after recanalization. A portion of
   the blood vessels was fixed with 4% paraformaldehyde and cut into
   frozen sections for immunofluorescence staining of CD31(1:100, abcam,
   ab281583), α-SMA (1:200, Boster, BM0002), ZO-1 (1:200, Proteintech,
   21773-1-AP) and eNOS (1:50, abcam, ab300071) to testify the morphology
   of ECs and its ability to generate NO.

Therapeutic efficacy in pig thrombus model

   Domestic pigs weighing 60–80 kg were on a diet 1 day before surgery.
   Before the operation, ketamine (6 mg/kg) and diazepam (0.2 mg/kg) were
   injected intramuscularly, and endotracheal intubation, oxygen, and
   sevoflurane were supplied after fixation. Venous cannula (18~20 g) was
   placed by ear vein puncture, and normal saline and propofol were
   intravenously drip. Under sterile conditions, the femoral artery was
   separated and rinsed with normal saline. Then the blood vessels were
   covered with filter paper strips (7 mm × 3 mm) that were soaked in 50%
   ferric chloride solution for 15 min, and the blood vessels were rinsed
   with normal saline after being left for 5 min^[275]46. Then,
   hmSi-CREKA-RB-PFH (15 mL, 20 mg/mL) was slowly injected through the ear
   vein, US (1.0 MHz, 1.5 W/cm^2) was performed for 30 min 1 h after
   injection. The size and blood flow of the thrombus were observed with
   Doppler ultrasound (VINNO 6VET/6LAB) every 10 min. Blood samples from
   the normal, thrombosis, and thrombolysis periods were collected for
   D-dimer detection, which was performed in the Laboratory Department of
   the Second Affiliated Hospital of Army Medical University Hospital.
   Finally, histological and immunofluorescence staining were conducted on
   the 20 μm frozen carotid artery sections.

Statistics and reproducibility

   The data are presented as the mean ± SEM. Statistical analyses were
   conducted using IBM SPSS Statistics 26 (IBM Software, Chicago, IL,
   USA). The Shapiro–Wilk test was employed to assess normality. The
   two-tailed paired samples t test was employed to compare the changes of
   the same sample at different stages, and the two-tailed independent
   samples t test was used for comparison between the two groups. Unless
   otherwise indicated, in the case of multiple groups, one-way analysis
   of variance (ANOVA) with the LSD post hoc test was utilized for
   parametric data, whereas Welch’s ANOVA with Tamhane’s T2 post hoc
   analysis was employed for non-parametric data. The specific statistical
   analyses conducted for each experiment are detailed in the
   corresponding figure legends. Statistical significance was set at
   P < 0.05.

Reporting summary

   Further information on research design is available in the [276]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [277]Supplementary Information^ (3.3MB, pdf)
   [278]Peer Review File^ (28.5MB, pdf)
   [279]41467_2024_50741_MOESM3_ESM.pdf^ (177.7KB, pdf)

   Description of Additional Supplementary Files
   [280]Supplementary Data 1^ (12.7KB, xlsx)
   [281]Supplementary Movie 1^ (44.5MB, mp4)
   [282]Supplementary Movie 2^ (5.1MB, mp4)
   [283]Supplementary Movie 3^ (991.4KB, mp4)
   [284]Supplementary Movie 4^ (1.7MB, mp4)
   [285]Reporting Summary^ (2.5MB, pdf)

Source data

   [286]Source Data^ (292.6KB, xlsx)

Acknowledgements