Ultrasound combined with SDF‐1α chemotactic microbubbles promotes stem cell homing in an osteoarthritis model

Abstract Osteoarthritis (OA) is a common joint disease in the middle and old age group with obvious cartilage damage, and the regeneration of cartilage is the key to alleviating or treating OA. In stem cell therapy, bone marrow stem cell (BMSC) has been confirmed to have cartilage regeneration ability. However, the role of stem cells in promoting articular cartilage regeneration is severely limited by their low homing rate. Stromal cell‐derived factor‐1α (SDF‐1α) plays a vital role in MSC migration and involves activation, mobilization, homing and retention. So, we aim to develop SDF‐1α‐loaded microbubbles MB(SDF‐1α), and to verify the migration of BMSCs with the effect of ultrasound combined with MB(SDF‐1α) in vitro and in vivo. The characteristics of microbubbles and the content of SDF‐1α were examined in vitro. To evaluate the effect of ultrasound combined with chemotactic microbubbles on stem cell migration, BMSCs were injected locally and intravenously into the knee joint of the OA model, and the markers of BMSCs in the cartilage were detected. We successfully prepared MB(SDF‐1α) through covalent bonding with impressive SDF‐1α loading efficacy loading content. In vitro study, ultrasound combined with MB(SDF‐1α) group can promote more stem cell migration with highest migrating cell counts, good cell viability and highest CXCR4 expression. In vivo experiment, more BMSCs surface markers presented in the ultrasound combined with MB(SDF‐1α) group with or without exogenous BMSCs administration. Hence, ultrasound combined with MB(SDF‐1α) could promote the homing of BMSCs to cartilage and provide a novel promising therapeutic approach for OA.


| INTRODUC TI ON
Osteoarthritis (OA) is a group of common joint diseases in the elderly, characterized by articular cartilage destruction, excessive joint stiffness, pain, and spasm. OA often involves multiple joints, mostly in the knee joint, elbow joint, hip joint, hand joint. According to the statistical analysis of the United Nations, by 2050, 130 million people worldwide are expected to develop the disease, and 40 million of them will develop into severe OA. 1 There is tremendous economic pressure in the treatment and management of OA, and the quality of life of patients can be significantly reduced.
The lesions of OA are mainly concentrated on articular cartilage. Articular cartilage is a transparent lining on the joint surface of the bone that cushions external impact and effectively reduces friction between the bones for smooth, painless joint movement. 2 The key to treating OA is to repair the damaged articular cartilage. However, treatment for articular cartilage defects still presents a major therapeutic challenge because of its hypocellular nature, vulnerability to injury and poor capacity to spontaneous regeneration.  3 The researchers used MSCs in the treatment of OA, with or without drugs or stents, and achieved a certain degree of efficacy. 4-10 All of the above studies have focused on the application of exogenous MSCs to local joints, but OA is a disease involving multiple joints, and it would be better if it could be treated systemically. In the current researches on systemic application of MSCs, the low homing rate of MSCs in target organs significantly affects the repair of target organs. Effective MSC therapy requires that MSCs reach the injured site. 11 So, increasing the number of stem cells in target organs, and promoting the homing to target organs is the key to promote tissue regeneration. 12,13 Stromal cell-derived factor-1 (SDF-1) is an extremely important chemokine for MSCs. It includes two subtypes, SDF-1α and SDF-1β, and SDF-1α plays a major role in chemotaxis. SDF-1 is well known for MSCs activation, mobilization, homing and also retention. 14 CXCR4 is a chemokine receptor that presents on the surface of stem cells.
SDF-1α and CXCR4 form the most important biological axis to promote homing of stem cells. [15][16][17] Increasing the expression of SDF-1α or CXCR4 can raise the homing efficiency of MSCs. 18,19 Microbubble (MB) is a type of monolayer bubble having a diameter between 0.5 and 10 μm. It is often used as a contrast agent and a targeted drug in medical imaging, 20 such as antibodies, polypeptides or transferrin vectors [21][22][23] and gene delivery vectors. 20 Ultrasoundtargeted microbubble destruction (UTMD) has been confirmed to increase the homing of transplanted MSCs to target organs, 24 and also provide a suitable environment for the survival of migrated stem cells. 25 In addition, UTMD can also promote the involvement of endocrine factors in stem cell homing. 24 Herein, in this study, we designed a chemotactic microbubble MB(SDF-1α) through the covalent bonding of SDF-1α to the microbubble shell, based on the perspective of SDF-1/CXCR4 axis regulation of stem cell homing. The chemotactic microbubbles are locally released SDF-1α to the target tissue under the irradiation of ultrasound. Next, BMSCs were injected intravenously, and pathological detection of BMSCs surface markers was performed for further detection and analysis. In this way, we aimed to potentiate the BMSCs homing to the cartilage in order to provide a promising approach for effective stem cell therapy for OA. Rat bone marrow stem cells (BMSC) were purchased from Cyagen Biotechnology Co., Ltd. Transwell chamber and six-well plate were purchased from Corning. APC-CXCR4 antibody was purchased from Thermo Fisher. Anti-rat-CXCR4, anti-rat-CD73 and anti-rat-CD90/ Thy 1 were purchased from Abcam. Perfluoropropane (SF 6 ) was purchased from Jietong Gas Technology Co., Ltd.

| Main Materials
Adult male Sprague-Dawley (SD) rats (weighed 200-250 g) were supplied by Chengdu Dashuo Experimental Animal Co., Ltd. In this experiment, all animal procedures were carried out in accordance with the protocol of the Animal Research Committee of Sichuan University.

| Preparation of MB and MB(SDF-1α)
The common MBs that not containing SDF-1α were basically prepared according to the methods of other researchers, 26 and some improvements of the protocol were also made. First, the composite membrane of DSPC and PA was prepared. Then mix the materials including DSPC + PA, DSPE-PEG2000-COOH, and PEG-4000 (mass ratio of 1:1:50) with ultrapure water. A 60°C water bath could help dissolve and mix thoroughly. The equal amount of the mixed solution is then dispensed into a vial and quickly lyophilized into a powder.
The powder was then dissolved by the solution that consists of 50% glucose, propylene glycol, and glycerine with volume ratio of 8:1:1.
After sealing the above-mixed lipid solution, the air in the bottle is removed by a gas suction device, and then the SF 6 is rapidly filled in.
Finally, place the vial in the shaker for oscillation for 45 seconds at a frequency of 4300/min and amplitude of 15 mm. According to this method, common MBs are obtained.
MB(SDF-1α) is synthesized by a covalent bond through the carboxyl group of the lipid shell and the amino group carried by SDF-1α.
Briefly, the carboxyl group of DSPE-PEG2000-COOH needed to be activated by EDC/NHS. That is, EDC (5 μL, 50 mg/mL) and NHS (5 μL, 100 mg/mL) were respectively added to the MB solution before gas replacement. After 15 minutes of activation, 20 μL SDF-1α solution containing 20 μg SDF-1α were added and another 2 hours were needed for the reaction. The resultant solution was treated following the same protocol as that used for common MBs to form the preliminary MB(SDF-1α) ( Figure 1A). Finally, the preliminary MB(SDF-1α) needed to be purified by centrifugation at 1000 rpm for 5 minutes to remove the free SDF-1α which was not loaded to the MBs, as well as the coupling agents. To evaluate the efficiency of conjunction between SDF-1α and MBs, we labelled SDF-1α with FITC prior to the experiment.

| Physical properties and characterization of MB(SDF-1α) and MB
The concentration and particle size of MB(SDF-1α) and MB were determined with Beckman Coulter (Multisizer 4e) at hours 2, 8 and 24, respectively. The morphology, size, and distribution of the two microbubbles were observed using an optical microscope (AX10 imager A2/AXP10 cam HRC; Zeiss). FITC fluorescence expression was observed using an inverted fluorescence microscope (OBSERVER D1/AX10 cam HRC; Zeiss) and flow cytometry to determine the conjunction rate of SDF-1α, respectively. In addition, enzyme-linked immune sorbent assay (ELISA) (R&D Systems, Inc) was applied to quantitatively determine the loaded SDF-1α. Finally, the SDF-1α encapsulation efficiency and loading content were calculated, respectively.

| Preparation of Bone marrow mesenchymal stem cells
Rat BMSCs were purchased from Cyagen Biotechnology Co., Ltd.
These cells have been identified as bone marrow MSCs by flow cytometry and adipogenic and osteogenic differentiation induction before the sale. All BMSCs were passed to the third generation and F I G U R E 1 Preparation of MB(SDF-1α) and OA animal model. A, Flow chart of chemotaxis microbubble MB(SDF-1α) preparation. The carboxyl group carried by the lipid molecule and the amino group of SDF-1 are linked by a covalent bond. After all, chemotactic microbubbles containing C3F8 and SDF-1α linked in the outer shell were prepared. B, The modified Hulth's modelling method for OA. During the operation, the medial meniscus is removed, and the anterior cruciate ligament is severed, causing joint instability and eventually osteoarthritis. An OA model was established on bilateral knees prepared for use. In subsequent experiments, the concentration of stem cells was adjusted according to the needs of the experiment.

| BMSCs migration experiment in vitro
This part of experiment included two sections, chemotaxis experiment, and chemotaxis inhibition experiment. In the chemotaxis experiment, adjust the BMSCs concentration to 2 × 10 5 /mL using the basal medium. All of the experiments were carried out in sixwell plates. The wells were divided into upper chambers and lower chambers. Add 1 mL BMSCs to all upper chambers respectively. And different experimental operations were performed on the lower chambers according to different groups. There are five groups in this study, control group, SDF-1α group, MB group, MB(SDF-1α) group and MB(SDF-1α) + ultrasound (US) group ( Table 1). The ultrasonic conditions of group five are duty cycle 20%, sound intensity 1 W/cm 2 , irradiation time 30 seconds, and frequency 1 MHz. After the operation of each group was completed, they were incubated at 37°C, 5% CO 2 for 8 hours. Then, the cells that migrated to the lower side of the transwell were stained with 0.1% crystal violet for 20 minutes and then washed with water. Finally, the cells were tested as follows: cell counts, cell activity assay using CCK-8 kit, flow cytometry to detect the existence of CXCR4 and qRT-PCR to determine the CXCR4 expression.
In the chemotaxis inhibition experiment, the key point was to incubate BMSCs with the CXCR4 receptor antagonist AMD3100 (5 μg/mL) for 30 minutes to block CXCR4. Then the group experiments were performed after adjusting the cell concentration. And finally, the numbers of migrated cells in each group were counted.

| Animal preparation
A total of 144 adult male SD rats (weighed 200-250 g) were selected to establish bilateral knee OA models by modified Hulth's modelling method ( Figure 1B). Briefly, rats were anesthetized by intraperitoneal injection of 0.3% sodium pentobarbital solution at a dose of 1 mL/100 g. The rats were placed supine on the operating table, and the area in front of the knee was prepared and disinfected using iodophor. Subsequently, the knee joint of the rat was kept at the maximum knee flexion, and the skin and the subcutaneous layer were longitudinally cut to fully expose the patellar tendon. Open the junction of the tendon and the muscle in the outer side of the patellar tendon, and cut it deeply into the joint cavity. The knee joint is then laid flat, then longitudinally cut along the medial side of the knee, and the patellar tendon was turned to the outside of the joint to expose the femoral trochlear. The medial meniscus of the knee joint was removed, and the anterior cruciate ligament was cut, while the posterior cruciate ligament was still preserved. A drawer test was performed on the knee joint of the rat to verify the stability of the joint. If the drawer test was positive, the operation was considered successful. All rats underwent bilateral knee arthroplasty and were housed in SPF animal laboratories. All experiments were performed according to the rules of the Animal Research Committee of Sichuan University.

| Grouping of in vivo experiments and experimental procedures
The in vivo study was divided into two main parts:  Table 2. After the OA models were established, rats were randomly assigned to each experimental group, 12 rats in each group. Except for the intra-articular injection group, both

| Pathological examination
The knee joint is inherently hard and cannot be sliced directly. It must be decalcified and then paraffin-embedded. When the prepared wax block is sliced, one slice is collected at intervals of five pieces, and the slice thickness is 4 μm. Each group of knee joints received haematoxylin and eosin staining (HE staining), Safranin O-Fast green staining and immunohistochemistry (IHC).

| Safranin O-Fast green staining
Safranin O is a basic dye that can be red in combination with acidic proteoglycans; fast green is an acid dye that combines with acidic collagen to appear green. The semi-quantitative analysis of collagen components in the cartilage matrix can be performed by the image of Safranin O-solid green staining, and then the modified Mankin's scoring standard is used to obtain the degeneration of cartilage tissue after comprehensive scoring. 28 It can visually reflect the structure of articular cartilage, subchondral bone, and bone tissue, and is superior to HE staining in display structure.

| Immunohistochemistry
Immunohistochemistry was performed to compare stem cell marker expression between different groups of rats. Finally, the target sequence was amplified. The amplification primers for CXCR4, CD70, and CD90 mRNA were displayed in Table 3.
The experiment was performed with a LightCyder 480 Real-Time PCR system (Roche). Then the relative expressions of mRNA were calculated. The results of the qRT-PCR were analysed using the pairwise fixed reallocation randomization test.

| Statistical analysis
IBM SPSS Statistics for Windows v22.0 (IBM Corp.) was used for data analysis. All continuous data were expressed as the mean ± standard deviation. One-way analysis of variance and Chi-square test were used for the statistical evaluation. The statistical significance level was set at P value < .05.

| Characterization of microbubbles
We have successfully prepared ordinary microbubbles MBs and chemotactic microbubbles MB(SDF-1α). The two kinds of microbubbles were similar on an unaided eye and presented as milky suspensions ( Figure 2). They were stable at room temperature and there was no statistical difference in the count and particle size of the microbubbles in a short period of time (Table 4).
Observed under the optical microscope, both MB(SDF-1α) and MB showed a spherical structure with uniform size and uniform distribution ( Figure 2). And both microbubbles showed irregular motion.
Under an inverted fluorescence microscope, the green fluorescence expressed by FITC was observed on the surface of MB(SDF-1α), while MB showed no green fluorescence ( Figure 2). Flow cytometry indicated that the carry rate of FITC fluorescence was greater than 80% ( Figure 2). The SDF-1α encapsulation efficiency was 76%, and the loading content was 15 μg/mL.

| BMSCs migration in vitro
In the chemotaxis experiment and chemotaxis inhibition experiment, the number of cells in the lower chamber of each experimental group was detected using a fully automatic cell counter.
We found that the number of migrated cells in SDF-1α group and MB(SDF-1α) + US group was higher than that in the other three

| IHC
In the experiment of verification of exogenous stem cell homing, the IHC results of CD73, CD90 and CXCR4 showed that the staining intensity of BMSCs + MB(SDF-1α) (I) + US group was higher than that of intravenous injection, and the staining intensity of BMSCs + MB(SDF-1α) + US group was higher than the other intravenous injection groups (P < .05; Figure 6A).

TA B L E 4 Beckman Coulter determination of concentration and particle size of microbubbles
While in the experiment of verification of endogenous stem cell homing, the IHC staining of CD73, CD90 and CXCR4 in each group was lower than that in the local injection group (P < .05). Except for the blank control group, the staining intensity of each group in verification of exogenous stem cell homing was higher than that in the endogenous stem cell groups ( Figure 6B).

| qRT-PCR
The expression levels of CXCR4, CD73 and CD90 mRNA in BMSCs + MB(SDF-1α) (IA) + US group were higher than those in the other groups in exogenous stem cell homing experiment (P < .05).
And CXCR4, CD73 mRNA expression level in MB(SDF-1α) (IA) + US group without exogenous BMSCs was higher than the other groups (P < .05; Figure 7), which indicated the possibility of endogenous BMSCs homing. The combination of ultrasound and microbubbles can cause sonoporation under proper conditions. With the effect of sonoporation, transient and non-lethal porosity will occur on biological membranes, which will increase the permeability of capillaries. 33,34 In addition, Meijering et al 35  This preliminary research still has several limitations. We have no traces of exogenous stem cells in this study. In addition, there is no further study on the long-term survival and differentiation of exogenous stem cells. This will be covered in our future research.
It has good stability and high unloading content and encapsulation ef-

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data were included in the manuscript.