Orthogonal test design for the optimization of superparamagnetic chitosan plasmid gelatin microspheres that promote vascularization of artificial bone.

Abstract The optimal conditions for the preparation of superparamagnetic chitosan plasmid (pReceiver‐M29‐VEGF165/DH5a) gelatin microspheres (SPCPGMs) were determined. Then, the performance of the SPCPGMs during neovascularization was evaluated in vivo. The SPCPGMs were prepared through a cross‐linking curing method and then filled into the hollow scaffold of an artificial bone. Neovascularization at the bone defect position was histologically examined in samples collected 2, 4, 6, and 8 weeks after the operation. The cellular magnetofection rate of superparamagnetic chitosan nanoparticles/plasmid (pReceiver‐M29‐VEGF165/DH5a) complexes reached 1–3% under static magnetic field (SMF). Meanwhile, the optimal conditions for SPCPGM fabrication were 20% Fe3O4 (w/v), 4 mg of plasmid, 5.3 mg of glutaraldehyde, and 500 rpm of emulsification rotate speed. Under oscillating magnetic fields (OMFs), 4–6 μg of plasmids was released daily for 21 days. Under the combined application of SMF and OMF, evident neovascularization occurred at the bone defect position 6 weeks after the operation. This result is expected to provide a new type of angiogenesis strategy for the research of bone tissue engineering.


| INTRODUCTION
Large segmental bone defect is one of the major problems that require urgent clinical solution (Verrier, Alini, Alsberg, et al., 2016;Yassine, Mokhtar, Houari, Karim, & Mohamed, 2017). Tissue engineering is an effective method for bone restoration, and angiogenesis in tissue-engineered bone is a key factor in the use of this technology in clinical applications (Almubarak et al., 2016;Fan, Crawford, & Xiao, 2010;Jin et al., 2016;Zhao et al., 2017). In vivo, cells that are more than 200 μm away from blood capillaries in artificial bones rarely survive because of insufficient nutrient and oxygen supplies (Lovett, Lee, Edwards, & Kaplan, 2009). Hence, the in vitro structure of a tissue-engineered bone is insufficiently capable of repairing large bone defects in most cases. Therefore, multiple artificial bone vascularization methods have been adopted, particularly the application of prevascularized engineered bone tissues and a graft combined with cocultured endothelial cells, osteoblasts, and stem cells (Almubarak et al., 2016;Fan, Zeng, Wang, Zhu, & Pei, 2014;Myeroff & Archdeacon, 2011;Zhang et al., 2016). However, this method has a limited clinical application because it requires complex operations and the availability of tissue materials with blood vessels is limited (Remulla et al., 1995). Vascular endothelial growth factor (VEGF) 165 is one of the most abundant and potent angiogenic agents among all VEGF isoforms (Otrock, Makarem, & Shamseddine, 2007;Street et al., 2002;Yang et al., 2014). However, it is deactivated in the presence of enzymes (Raftery, Mencía, Chen, et al., 2017). Given the progress on the research of control release techniques, the slow release of active growth factors is attracting increased attention (Dionigi et al., 2014;Xie et al., 2016). For example, VEGF encapsulated with calcium alginate beads through the extrusion/external gelation method has a constant release rate of 6 ng/ml/day and may be sustained for 14 days (Gu, Amsden, & Neufeld, 2004). A system wherein the controlled release of insulin is regulated by applying oscillating magnetic field (OMF) increased the release of insulin at levels threefold of that of the control group (Finotelli, Da, Sola-Penna, et al., 2010). Surface cationic magnetic chitosan-iron oxide nanoparticles can potentially enhance magnetofection efficiency under static magnetic field (SMF; Sohrabijam, Saeidifar, & Zamanian, 2017). A porous nano-hydroxyapatite/polyamide66 (n-HA/PA66) composite has been developed in recent years. Yan proved that the porous n-HA/PA66 composite is biologically safe and exhibits good biocompatibility, osteoinduction, and osseointegration (Xiong et al., 2014). Meanwhile, the use of a magnetic carrier drug microsphere and bioactive artificial bone under magnetic fields in vitro (OMF and/or SMF) for vascularization has not been reported before.
Basing on those above theories, we constructed superparamagnetic chitosan plasmid (pReceiver-M29-VEGF165/DH5a) gelatin microspheres (SPCPGMs) and obtained their optimal formula through a cross-linking curing method. Then, we poured the SPCPGMs into a porous n-HA/PA66 scaffold. The filled scaffold was planted in a model of a large segmental radius bone defect in a New Zealand rabbit. The in vitro release of the plasmids in the SPCPGMs was observed in the presence of OMF. in vivo vascularization in the artificial bone was also observed. This application is expected to provide a new type of angiogenesis strategy for bone tissue engineering.

| Preparation and characterization of SPCN
Superparamagnetic chitosan nanoparticles (SPCNs) were prepared through a chemical coprecipitation method (Guo, Liu, Hong, & Li, 2010). Specifically, a mixture containing ammonium ferrous sulfate and ammonium ferric sulfate was dissolved with 2% chitosan (w/v) in acetic acid solution (pH 5.5). The resulting solution was designated as Solution A. The solution was transferred into a 500 ml threenecked flask. For Solution B, 6 mol/L NaOH was prepared. Under the protection of a nitrogen atmosphere, Solution A was heated to 55 C with vigorous stirring. Solution B was added dropwise into Solution A for 10 min. The stirring speed was lowered, and the reaction mixture was heated to 85 C for 90 min. An appropriate amount of glutaraldehyde was added to react for 30 min. The pH of the resulting suspension was adjusted to 5.5 with diluted hydrochloric acid. The obtained black colloidal precipitate was washed with distilled water and dried under vacuum to form a dry powder, which was stored at room temperature. The morphologies of the prepared SPCNs were observed by TEM (Hitachi-600, Japan), and the diameter and distribution were measured by using a laser diffraction particle size analyzer (Rise-2008, China). The electric potential was measured by a zeta potential analyzer (Malvern, UK). Nitrogen content in chitosan was determined using a Kjeldahl apparatus (QSY-2, China). The structural characteristics of the samples were evaluated using FT-IR (Shimadzu, Japan).

| In vitro cell transfection of SPCPNC under SMF
VEGF165 plasmids (pReceiver-M29-VEGF165/DH5a) were amplified in large quantities, and a competent-state DH5α Escherichia coli strain was used. The plasmids were extracted, purified, and concentrated with a plasmid extraction kit (OMEGA, GA, USA). We used a SmartSpecTM3000 nucleic acid protein determinator (Bio-Rad, California, USA)) to determine the concentration of nucleic acid.

| Optimized preparation and characterization of SPCPGM under orthogonal design
An orthogonal design was used for the optimization of the parameters for SPCPGM preparation. The cross-linking curing method was then used for the preparation of SPCPGM with different components. Days of microsphere plasmid release and saturation magnetization were considered comprehensive evaluation indexes. An orthogonal test was conducted according to L9 (34) ( Table 1). CPGM was used as control. The procedures of this method were as follows: magnetic chitosan nanoparticles containing different proportions of Fe 3 O 4 were added to 5 ml of acetic acid buffer (pH 5.5). The resulting mixtures were heated to 55 C. A certain dose of plasmid (pReceiver-VEGF165/ DH5a) was heated to 55 C and then mixed by vortexing for 20 s and subjected to a binding reaction for 1 hr. Approximately 20% of the gelatin solution (w/v) was preheated at 50 C, then added and mixed to form a composite emulsion, which was subsequently added to paraffin oil (50 C) and preheated with a homogenizer (IKA-T25, Germany) at a proper rotation speed. After confirming the presence of microspheres, we lowered the temperature to 4 C with an ice bath.
An appropriate amount of 37% formaldehyde and isopropanol was added, and the resulting mixture was continuously stirred for 1 hr.
The samples were washed from five to eight times with ether and deionized water (the washing solution was saved to determine its plasmid content). The microspheres were collected by centrifugation, lyophilized, and stored at −20 C. Superparamagnetic chitosan was replaced with chitosan for the preparation of CPGM.
The magnetic properties of SPCPGM were measured with a BHV-55 vibrating sample magnetometer (VSM). The morphologies were then investigated by light microscopy (LM; Olympus-CK-2, Japan) and scanning electron microscopy (SEM; Hitachi-S-3000N, Japan). The diameter and distribution were measured by using a laser diffraction particle size analyzer (Rise-2008, China). Three batches of microspheres were prepared according to the optimized formula. Scrubbing solutions from all groups were collected and centrifuged (8,000 rpm, 5 min). We collected the supernatants and used the UV method to determine pReceiver-VEGF165/DH5a plasmid concentration at a 260 nm wavelength (we used a microsphere scrubbing solution without plasmids to set the equipment at zero value and to eliminate interference from auxiliary materials). The weights of these microspheres were measured. The drug-loading capacity and encapsulation efficiency of microcapsules were calculated according to the following formula: encapsulation effi- We selected 32 New Zealand white rabbits to establish a bone defect model at two sides of a rabbit radius. These rabbits were randomly divided into four groups as follows: Group A: SPCPGM + SMF + OMF;

| In vitro plasmid release experiment of SPCPGM under OMF
Group B: SPCPGM + SMF; Group C: single application of SPCPGM; and group D: CPGM (blank control). At the 2nd, 4th, 6th, and 8th week after the operation, Su Mian Xin (0.2-0.3 ml/kg) was used as intramuscular anesthetic for the animals. The total radioactive count and average count value were first determined in the bone scaffold region through 99MTc-MDP radionuclide tomography and then used for analyzing vascularization at the bone scaffold. The blood vessels were perfused and dyed with ink. Finally, the animals were killed by injecting air into their ear veins.
The samples were collected, and the status of angiogenesis in the scaffold was observed. Paraformaldehyde (4%) was used to fix the samples, which were then prepared and dyed with HE. Qualitative and preliminary observation of the vascularization was performed under an optical microscope.
Newly generated blood vessels were determined qualitatively and quantitatively by using the stereological image generated by the Tiger 920G image analysis software system. The conditions of the newly generated vessels in the four groups, each of which were subjected to different treatments, were evaluated through statistical analysis.

| Statistical analysis
SPSS 10.0 software package was used for statistical analysis and data processing. All data were expressed as mean ± standard deviation.
Variance analysis (one-way ANOVA) was used for multigroup comparisons. Statistical analysis of the obtained data was implemented.
p < .05 indicated that the difference is statistically significant while p < .01 indicated that the difference is highly significant.

| Preparing and characterizing SPCNs
SPCNs were prepared by coprecipitation (Guo et al., 2010). The SPCNs were circular or oval and had favorable evenness, as observed T A B L E 1 Factors and levels of the orthogonal design through transmission electron microscopy (TEM; Figure 1A,B). Figure 1C shows the Fe 3 O 4 magnetic fluid, and Figure 1D shows the directional migration of magnetic fluids under a magnetic field.
The average particle size of SPCN was 0.046 ± 0.024 μm ( Figure 1E), the electric potential was positive, and the zeta potential on the SPCN surface was 70.5 ± 11 mV. The nitrogen content of SPCN was 0.007 mg/L. Figure 1F shows the Fourier transform infrared spectroscopy (FT-IR) spectra of chitosan ( Figure 1F-

| Preparation and in vitro magnetofection of the SPCNPCs
The results of AGE analysis and identification showed that the increase in nitrogen/phosphorus ratio (N/P) gradually weakened the brightness of the specific bright strips of the plasmids (Figure 2a).
Thus, plasmids escaping from the spotting holes gradually decreased.
At the N/P ratio of 1/0.5 (electrophoresis channel 11), no specific bright strip occurred, indicating that the SPCN had completely bound with the plasmids and were thus unable to escape from the spotting holes. The zeta potential of SPCNPC was 27.8 ± 4 mV. These results laid a foundation for the subsequent in vitro transfection experiment.   (w/v), 4 mg (w/v), and 5.3 mg (w/v), respectively. The emulsification rotation speed was 500 rpm.

| VERIFICATION OF OPTIMIZED PRESCRIPTION
Three batches of SPCPGM samples were prepared by using the optimal results shown in Table 1 and observed through LM and SEM.
Most of the microspheres were round and spherical and had concave-convex surfaces, uniform size, and favorable dispersity (Figure 3a,b). The average particle size was 65.358 ± 20.931 μm ( Figure 3c). Figure 3d shows the hysteresis loop of SPCPGM at room temperature. The magnetization of the samples approached saturation values when the applied magnetic field was increased to 25,000 Oe.
The saturation magnetization of SPCPGM was 8.223 emu/g. A small remnant magnetization of 0.0199 emu/g was obtained at the external magnetic field of 0 Oe, indicating that the magnetic particles produced were superparamagnetic. The results show that the encapsulation efficiency and drug-loading capacity of SPCPGM were 98.51% ± 1.34% (w/w) and 53.42% ± 1.54% (w/w), respectively.
OMF was used in the in vitro plasmid release experiment for the SPCPGMs. OMF was applied to group A because the daily drug release was~4-6 μg at the 1st day of the experiment. The sustainable release quantity of Group A was apparently higher than those in the other two groups (~four times). Results of the statistical variance analysis showed that Group A was significantly different from Groups B and C in terms of drug dissolving-out quantity (Q test, p < .001). The difference between Groups B and C in drug-release quantity was nonsignificant (Q test, p > .05).
The dissolution rate of Group A from the 22nd to the 25th days of the experiment slowed down and drug-release percentage of the plasmids was~70%. The approximate values of the drug-release percentage of the plasmids in the groups not subjected to OMF were 11% in SPCPGM and 15% in chitosan plasmid (pReceiver-M29-VEGF165/DH5a) gelatin microspheres (CPGMs; Figure 3e).

| IN VITRO EXPERIMENT OF VASCULARIZATION OF ARTIFICIAL BONE
An artificial bone scaffold in a hollow structure with lateral holes was prepared by using n-HA/PA66 bone cement with good histocompatibility ( Figure 4a). Then, the SPCPGMs was loaded into the hollow portion of the scaffold. A model of a large segmental radius bone defect was established in New Zealand rabbits (Hou et al., 2015;Figure 4b), and an artificial bone scaffold loaded with SPCPGM was implanted.
Lastly, the angiogenesis results of the artificial bone scaffold were observed under different magnetic fields.

| OBSERVATION OF GROSS MORPHOLOGY
On the 1st day after the operation, experimental animals started eat- On the 2nd week, the implants in Group A were wrapped by loose fiber tissues. A small quantity of vascularized soft tissues grew into the side holes of the implants. One-third of the microspheres formed residues in the hollow parts of the implants. In Group B, the implants were wrapped by loose fiber tissues without obvious neovascularization. In Groups C and D, half of the microspheres formed residues in the hollow parts of the implants wrapped by a small amount of fiber tissues. The residues of the microspheres in the hollow parts of the implants was~1/2.
On the 4th week, the implants in Group A were wrapped by fiber tissues. A large amount of vascularized soft tissues grew into the side holes of the implants, and a small amount of microsphere residue was observed in the hollow part. In Group B, a small amount of vascularized soft tissues grew into the side holes of the implants, and the microsphere residue in the hollow part was~1/3. In Groups C and D, a small amount of vascularized soft tissues grew into the side holes of the implants and microsphere residue in the hollow part of the implants was~1/3.
On the 6th week, the implants in Group A were completely wrapped by compact fiber membranous tissues and a large quantity of vascularized soft tissues grew into the side holes of the implants (Figure 4c). In Group B, vascularized soft tissues grew into the side holes of the implants. Meanwhile, a large amount of vascularized soft tissues grew in the side holes of the implants in Groups C and D.
On the 8th week, the implants in Group A were completely wrapped by compact fiber membranous tissues, and a large amount of vascularized soft tissues grew into the side holes of the implants.
F I G U R E 5 In vivo experiment of SPCPGM facilitating artificial bone vascularization under magnetic fields (OMF and/or SMF). a, Ink dyeing for rabbit radius implant (×40). S is the n-HA/PA66 artificial bone scaffold. The red arrow shows the blood vessel dyed after ink perfusion. b, Hematoxylin and eosin staining (×40). S is the n-HA/PA66 artificial bone scaffold. The green asterisk represents residual microspheres. The black arrow points at new vessels. c, Variance analysis of ink dyeing after operation. The data are expressed as the means ± standard deviation (SD) of 64. ** p < .001. d, Radionuclide tomography.
The pictures show a collection of radionuclides of Group A in the sixth week. e, Average radioactive count of radionuclide blood flow phase. The data are expressed as the means ± SD of 64. ** p < .001 Regenerated tissues were also dyed into black after ink perfusion. In Group B, vascularized soft tissues grew into the implant side holes.
Similarly, vascularized soft tissues grew into the side holes of the implants in Groups C and D but in large amounts. Figure 4d shows the schematic diagram of the SPCPGM mechanism involved in artificial bone scaffold vascularization. Superparamagnetic nanoparticles subjected to magnetic fields generated magnetic micromotion, which facilitated the local interchange of nutrients and entry of nutrients, such as protein, and oxygen into the scaffold while discharging metabolite. SPCPGM subjected to OMF released positively charged chitosan plasmid compounds, which adhered to the negatively charged cytomembrane enter before entering the cells. Angiogenesis was facilitated through the transcription, translation, and expression of the VEGF protein.

| HISTOLOGICAL OBSERVATION
The samples were dyed with ink ( Figure 5a) and hematoxylin and eosin (HE; Figure 5b). In Group A, the connective tissues grew around the implants and into the lumens and were infiltrated by a number of inflammatory cells 2 weeks after the operation. Immature blood capillaries were also observed. The ink-perfused blood vessels formed net-like structures. Few soft cartilages grew inward, and microsphere residues were occasionally observed. In the 6th week, mature blood vessels formed and fibrous porosis occurred inside the scaffold.
Nearly no microsphere residue was observed. In the 8th week, mature vascular nets and soft bones filled the lumen. Vascularization in Group B was similar to that in Group A but occurred 2 weeks later than that in Group A. A large amount of microsphere residues was observed in the 8th week. Vascularization times in Groups D and C were longer than that of Group B, and large nondegraded microsphere residues were observed. Figure

| DISCUSSION
The three basic processes after bone transplantation are implant vascularization, osteogenesis, and epiphysis fusion (Gross, Cox, & Jinnah, 1993). The key link is vascularization, which occurs during transplantation and restoration. Notably, vascularization remains to be the foundation of bone defect restoration after the transplantation of bioactive artificial bone (Almubarak et al., 2016). Sufficient blood supply is a crucial factor for the successful in vivo transplantation of bioactive artificial bone tissues. The dispersion and permeation of nutritive substances can only reach 150-200 μm around blood vessels (Lovett et al., 2009). Therefore, determining the possibility of rapid blood supply reconstruction after artificial bone transplantation remains a challenge when bioactive artificial bone research is applied to clinical settings.
Using all kinds of cell growth factors and combining them with scaffold materials for angiogenesis are among the main reconstruction strategies for supplying blood in artificial bones (Lindhorst, Tavassol, von, et al., 2010;Lovett et al., 2009;Sun et al., 2013). The other main strategies include using transgenic cells to construct bone tissues (Kawai et al., 2017;Kawai, Bessho, Maruyama, Miyazaki, & Yamamoto, 2006), using tissues that contain abundant vascular nets and wrapping or implanting them into a bone scaffold material (Li & Kawashita, 2011;Türer & Önger, 2017;Wu et al., 2018), using a drug (gene) release system for the vascularization of artificial bones or bone tissues, and performing vascularized bone tissue preconstruction (Hall, 2007;Lan, Tian, ZhuGe, et al., 2017;Moncion, Lin, O'Neill, et al., 2017). Bioactive artificial bones containing magnetic drug-carrying microspheres that facilitate vascularization under in vitro magnetic field (SMF or OMF) is currently unreported.
Therefore, we propose the application of an in vitro noninvasive OMF and SMF for vascularization. This approach promotes the micromovement of magnetic gene-loaded microspheres inside an artificial bone scaffold. SPCPGMs were constructed, and the optimal formula was obtained through a cross-linking curing method. The porous n-HA/PA66 scaffold was then filled with SPCPGM, and the resulting complex was planted in large segmental radius bone defects in New Zealand rabbits. The in vitro release of the plasmid in SPCPGM was observed under OMF.
According to the retrieved documents (Denkbas, Kilicay, Birlikseven, et al., 2002;Guo et al., 2010), the following data were obtained from magnetic microsphere preparation: gelatin proportion of~25%, temperature of 55 C, which decreased to 4 C after an ice
Daily release of plasmids by optimized SPCPGM under OMF was 4-6 μg for 3 weeks. These conditions are conducive to vascularization inside artificial bone scaffolds. The in vivo animal experiment confirmed that the joint action of SMF and OMF improved the effect of optimized SPCPGM on vascularization inside the scaffold, especially on the 6th week. Therefore, magnetic micromotion and interchange of nutritive substances inside a large segmental artificial bone scaffold can form a highly efficient system that enables the interchange of nutritive substances, magnetic micromotion, genetic release, and genetic transfer under external magnetic fields.
This system is expected to become a new strategy for the vascularization of various in vivo implants. The system enables the volume of artificial tissue engineering implants to reach the standard size range for clinical applications. We surmised that this method can be extensively used in restoring defects and treating large segmental bone tissues.

DISCLOSURE
No potential conflict of interest was reported by the authors.