Biotinylated, lipid-shelled MBs were prepared as previously described (Yan et al., 2013). Briefly, DSPC, DSPE–PEG2000 and DSPE–PEG2000-Biotin (Avanti Polar Lipids, Alabaster, AL, USA) (molar ratios = 9:0.5:0.5) were blended in chloroform. The solvent was evaporated under nitrogen flow at room temperature, forming a thin film on the wall of the test tube. Residual chloroform was further removed by vacuum treatment for at least 2 h. The completely dried phospholipid membrane was hydrated at 60°C with a given buffer consisting of 0.1 M Tris (pH 7.4):glycerol:propylene glycol (80:10:10 by volume), and then transferred into vials. After sealing the vials, the air in the vials was replaced with perfluoropropane (C₃F₈). Biotinylated MBs (MB_biotin) were obtained by mechanically vibrating for 30 s. The resulting MB_biotin were washed with phosphate buffered saline (PBS) solution twice to eliminate excess unincorporated lipids by centrifuge at 400 g for 3 min. Three microgram of avidin per 10⁷ MBs was then incubated with the MB_biotin dispersion for 30 min at room temperature, followed by rinsing three times to remove unreacted avidin (Aladdin, Pudong, Shanghai, China). Then either 2.5 μg of biotinylated goat anti-rat Type IV Collagen antibody or biotinylated goat IgG isotype control antibody (SouthernBiotech, Birmingham, AL, USA) was added to per 10⁸ MBs. After incubation for 30 min at room temperature, MB_ColIV or isotype control MBs (MB_Ctrl) were collected by centrifugation. Non-targeted MBs (MB_N) without conjugated primary antibody were also prepared.

To confirm the conjugation of Type IV Collagen antibody to the surface of MBs, Alexa Fluor® 555-labeled anti-goat secondary antibody (Invitrogen, Carlsbad, CA, USA) was added to MB_ColIV and mixed for 30 min at room temperature. The mixture was separated by centrifugation and the upper layer was washed twice with PBS to eliminate the free secondary antibody. The fluorescence-labeled MB_ColIV were examined by a fluorescence inversion microscopy (Olympus Corporation, Tokyo, Japan).

To analyze the characteristics of the MB_ColIV, MB_N were used as the blank control. The morphology of the MBs were examined by using bright-field and fluorescence inversion microscopy (Olympus Corporation, Tokyo, Japan). Particle size, size distribution and concentration of MBs were determined by using the Accusizer 780 Optical Particle Sizer (Particle Sizing Systems, Santa Barbara, CA, USA) with a 0.5 μm diameter detection limit.

All animal procedures were performed in accordance with the principles outlined in the Guide for the Care and Use of Laboratory Animals and approved by Animal Care and Use Committee. Adult male Sprague–Dawley rats (Vital River Laboratory, Beijing, China), weighing 400–500 g, underwent the rat carotid artery balloon injury as previously described (Chan et al., 2011; Petrasheskaya et al., 2016). Rats were given aspirin by oral gavage before surgery and kept anesthetized by isoflurane at 2% at 2 L/min oxygen. Following a midline neck incision, the left common carotid artery (LCCA), left external carotid artery (LECA), and left internal carotid artery (LICA) were exposed. After distal ligation of the LECA, the proximal end of LCCA and the distal end of LICA were then temporarily clipped by arterial clamps and a 2-French arterial embolectomy catheter (Edwards Lifesciences, Irvine, CA, USA) was advanced proximally into LCCA through the incision of LECA. The balloon was inflated and withdrawn with a rotating action to the arteriotomy in LECA and then reintroduced and retracted twice again. The balloon catheter was removed and the LECA was ligated. Finally, arterial clamps were removed to restore the blood flow, and then the neck incision was closed. Thereafter, rat carotid artery injury model was performed and used for the following experiments.

Bilateral carotid arteries were harvested following in situ perfusion fixation with PBS and 4% paraformaldehyde. Tissues were placed in paraformaldehyde for 1 h at 4°C, then overnight in 30% sucrose in PBS at 4°C for cryo-protection. Vessels were coated with Optimum Cutting Temperature O.C.T.™ compound (Tissue Tek, Hatfield, PA) and transferred to liquid nitrogen for flash-freeze. Vessels were sectioned to obtain arterial cross-sections across the length of the artery by cryostat microtome (CM1950; Leica, Heidelberg, Germany) and were examined histologically using routine hematoxylin-eosin (H&E) staining for morphometric analysis. Digital images were collected with bright-field by inversion microscopy (Olympus Corporation, Tokyo, Japan).

Attachment ability of MBs to collagen IV was evaluated by counting the number of adhering MBs. Briefly, after the rat carotid artery injury model and the in situ perfusion fixation were completed, the LCCA was harvested and then opened longitudinally. The side of adventitia was adhered to a coverslip. Then MBs were added into a 6-well plate (2 × 10⁷ MBs per well) and filled with PBS. Due to the static flotation nature of MBs, the coverslip with the artery coated faced downward to maximize interaction between the tissue and MBs for 5 min at room temperature. After that, the free MBs were removed by rinsing five times with PBS. To further determine the specificity of MBs adhesion, vessel tissues were pretreated with 25 μg/mL goat anti-rat Type IV Collagen antibody to block available receptors prior to MB_ColIV incubation. Digital images were obtained with bright-field by inversion microscopy (Olympus Corporation, Tokyo, Japan) to count the number of attached MBs in five random fields of view.

After rat carotid artery injury model was developed as described above, rats were kept anesthetized on a heated stage throughout the imaging session. Ultrasound molecular imaging was performed with a commercial ultrasound system Resona 7 (Mindray Medical Systems, Shenzhen, China) using a L11-3 linear array transducer. Contrast imaging mode was applied in the ultrasound molecular imaging experiments. All imaging parameters were kept constant throughout the whole procedure as follows: frequency 5.6 MHz, depth 2 cm, gain 45 dB, frame rate 10 Hz, dynamic range 115 dB, and mechanical index 0.085. In order to reduce motion interference, both the transducer and the rats were fixed to maintain the same long axis cross section of carotid artery. The concentration of MBs suspension was adjusted to 2 × 10⁸ MBs/ml. Then 200 μl MBs suspension was injected intravenously through tail vein following by flushing with 50 μl PBS. Four minutes after MBs injection, 100 frames of images were captured to obtain a signal from adherent and freely circulating MBs. A continuous high-power destructive pulse (mechanical index: 0.553) was then applied for 2 s to destroy these MBs. After 2 s, to allow the freely circulating MBs to replenish, another 100 frames of images were acquired, in which the ultrasonic signals were from any residual freely circulating MBs and tissue. To minimize the bias and test the specificity of these molecular imaging signals only resulting from adherent targeted MBs, MB_ColIV, and MB_Ctrl were administered in random order to all rats. Another 30-min delay was allowed to clear MBs from the preceding imaging session. As a control, ultrasound molecular imaging was also performed in normal rats to further assess the specificity of MBs adhesion. The acoustic imaging signals were analyzed by using commercially available analysis software (Mindray Medical Systems, Shenzhen, China). As previously described (Wu et al., 2011), the difference in signal intensity from adherent MBs was calculated by subtracting the post-destruction signal from the pre-destruction signal.

Immunostaining analysis for collagen IV was performed to confirm that collagen IV was exposed through endothelium denudation. Briefly, LCCAs from normal artery group and balloon-injured artery group were excised, followed by in situ perfusion fixation with PBS and 4% paraformaldehyde. Then artery samples were transferred to paraformaldehyde overnight, followed by ethanol dehydration and paraffin embedment. Artery samples were sectioned to obtain arterial cross-sections. After the process of deparaffinage and rehydration, slides were blocked and then incubated with the primary antibody goat anti-rat Type IV Collagen (SouthernBiotech, Birmingham, AL, USA), followed by being rinsed with PBS thrice and then incubated with horseradish peroxidase-conjugated secondary antibodies. Finally, slides were incubated with diaminobenzidine for horseradish peroxidase reaction. In order to exhibit the tissue morphology of vascular wall, slides were counterstained with hematoxylin. Immunohistochemical images were acquired by using inversion microscopy (Olympus Corporation, Tokyo, Japan) under bright field.

Data were analyzed by the statistic software SPSS 20.0 (IBM Corp, Armonk, New York, USA). As values were normally distributed, statistic difference between two groups was determined by Student's t-test, and one-way ANOVA were used to compare among more than two groups. If the differences among these groups were significant for one-way ANOVA, the difference between the two groups was conducted by Student-Newman-Keuls test. The P < 0.05 indicated statistically significant difference.

The fabricated MB_ColIV appeared a typical multi-peak particle distribution. There was no significant difference between MB_ColIV and MB_N for the particle size and distribution (all P > 0.05). The mean diameter of MB_ColIV and MB_N were 1.92 and 1.89 μm, respectively. The results demonstrated that the conjunction of avidin and biotinylated antibody to MBs surface did not change the particle size and distribution of MBs (Figure 2A). Figure 2B showed that both MB_ColIV and MB_N had smooth and spherical morphology for microscopic observation. The surface of MB_ColIV emitted a bright red fluorescence, indicating the anti-rat Type IV Collagen antibodies were successfully conjugated onto the surface of these MBs (Figure 2B).

Vascular balloon injury is a standardized protocol for endothelial denudation and intimal damage through repeated insertion and retreat of the balloon catheter (Figure 3A). In order to confirm the successful development of carotid artery injury model, cross-sections were stained with H&E. From the Figures 3B,C we can see that LCCA lost an endothelial monolayer after vascular balloon injury, compared with the normal right common carotid artery (RCCA). The results showed that we successfully constructed the carotid injury model.

To determine the binding specificity of MB_ColIV, we performed the binding specificity experiments ex vivo through incubating the MB_ColIV or MB_Ctrl with the injured LCCA. After a static exposure of the injured LCCA to MBs, we could see that a large number of MB_ColIV were adhered to the injured vessel (Figure 4A), which was significantly higher than MB_Ctrl (Figure 4B). Meanwhile, pre-blocking receptors in injured LCCA by anti-rat Type IV Collagen antibodies greatly reduced the attachment of MB_ColIV (Figure 4C). Quantitative analysis indicated that the binding efficiency of MB_ColIV was about four times higher compared with MB_Ctrl (282.1 ± 77.2 MBs per field vs. 71.07 ± 62.07 per field, P < 0.01) (Figure 4D).

To further evaluate the special adherence of MBs and their ultrasound imaging effect in vivo, ultrasound molecular imaging was performed in the balloon-injured carotid artery group (n = 8) and the normal carotid artery group (n = 7). As shown in Figure 5A, 4 min after MBs injection, almost all of MBs had been washed out. Meanwhile, MB_ColIV were hardly attached to the normal artery wall (Figure 5A). By contrast, MB_ColIV were obviously adhered to the inner wall of balloon-injured artery (Figure 5A and Supplementary Video 1). After the trigger of the destruction pulse, no echo signal of MBs was found in balloon-injured the carotid artery group and the normal carotid artery group (Figure 5A). This procedure further confirmed that MB_ColIV were attached to the inner vessel wall. Quantitative analysis revealed that the signal intensity of MB_ColIV (2.30 ± 0.91 dB) was higher than MB_Ctrl (0.57 ± 0.26 dB) in balloon-injured carotid artery group (P < 0.01) (Figure 5B). Moreover, the signal intensity of MB_ColIV in the balloon-injured carotid artery group was also significantly higher than normal carotid artery group (2.30 ± 0.91 dB vs. 0.31 ± 0.21 dB, P < 0.01) (Figure 5B).

To confirm the results of ultrasound molecular imaging, LCCAs of the normal group and balloon-injured group were harvested and analyzed for the exposure of collagen IV after endothelium denudation. The results of immunohistochemical staining examination were presented in Figure 6. We could see that both balloon-injured artery and normal artery were stained dark brown for collagen IV in basement membrane, indicating collagen IV was a major constituent of the basement membrane. It was consistent with the previous studies (Kalluri, 2003; Abreu-Velez and Howard, 2012; Manon-Jensen et al., 2016). In normal artery intima, collagen IV presenting in basement membrane, was covered by a monolayer endothelium (Figure 6A). However, without endothelium coverage, collagen IV was directly exposed in balloon-injured artery (Figure 6B). The exposure of collagen IV greatly enhanced the interaction opportunity with MB_ColIV. In addition, collagen IV was also a component of the basement membrane for SMCs in tunica media, which had been reported in previous studies (Shekhonin et al., 1985; Howard and Macarak, 1989).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.