Nanotextured silk fibroin/hydroxyapatite biomimetic bilayer tough structure regulated osteogenic/chondrogenic differentiation of mesenchymal stem cells for osteochondral repair

Abstract Objectives Articular cartilage plays a vital role in bearing and buffering. Injured cartilage and subchondral bone repair is a crucial challenge in cartilage tissue engineering due to the peculiar structure of osteochondral unit and the requirement of osteogenic/chondrogenic bi‐directional differentiation. Based on the bionics principle, a nanotextured silk fibroin (SF)‐chondroitin sulphate (CS)/hydroxyapatite (HAp) nanowire tough bilayer structure was prepared for osteochondral repair. Methods The SF‐CS/HAp membrane was constructed by alcohol‐induced β‐sheet formation serving as the physical crosslink. Its osteochondral repairing capacity was evaluated by culturing bone marrow mesenchymal stem cells (BMSCs) in vitro and constructing a rat osteochondral defect model in vivo. Results The bilayer SF‐CS/HAp membrane with satisfactory mechanical properties similar to natural cartilage imitated the natural osteochondral unit structural layers and exerted the function of bearing and buffering timely after in vivo implantation. SF‐CS layer upregulated the expression of chondrogenesis‐related genes of BMSCs by surface nanotopography and sustained release CS. Meanwhile, nanotextured HAp layer assembled with nanowire endowed the membrane with an osteogenic differentiation tendency for BMSCs. In vivo results proved that the biomimetic bilayer structure dramatically promoted new cartilage formation and subchondral bone remodelling for osteochondral defect model after implantation. Conclusions The SF‐CS/HAp biomimetic bilayer membrane provides a promising strategy for precise osteochondral repair.


| INTRODUC TI ON
Articular cartilage possesses the function of bearing and buffering and is an essential foundation for human body motion. 1 The peculiar structure of cartilage tissue, including deficiency of blood vessels and nerve, few chondrocytes and low proliferation ability, determines the poor self-repair capability after injury. Clinically, treatment of articular cartilage injury has always been an intractable problem owing to the weak self-repair ability of cartilage. 2,3 Currently, various cartilage repairing materials have been prepared from a variety of natural macromolecular compounds, such as gelatin, collagen, hyaluronic acid and chitosan. [4][5][6][7][8][9][10] It is worth noting that cartilage defect was usually companied by subchondral bone injury. Thus, the hierarchical scaffolds imitating the structure and function of cartilage and subchondral bone have been designed for the bionic repair of the integrated osteochondral unit. [11][12][13] Some hierarchical materials have achieved excellent efficacies in the restoration of osteochondral defect, but certain limitations existed, especially in mechanical performance. Once implanted into the arthrosis, many materials had weak mechanical strength to support heavy loads and stress in time. Therefore, besides biocompatibility and biological effects, suitable mechanical performance is also critical for osteochondral repair. 2 Furthermore, the hierarchical scaffolds, which used one material loaded with various chondrogenic and osteogenic inducers in two layers, could not truly simulate osteochondral repair environment. A few studies have developed biphasic scaffolds using different materials, such as the electrospinning silk/bioactive glass composite scaffolds and graphene-polycaprolactone (PCL)/bioactive glass scaffolds, to mimic the hierarchical complexity of the osteochondral interface. 14,15 However, owing to the absence of continuous phases in different layers, the connectivity between the two layers was insufficient. These limitations become a hindrance to clinical transformation. Accordingly, it is essential to develop a seamless bilayer structure with preferable mechanical property to achieve precise osteochondral regeneration.
Hydroxyapatite (HAp) has the superiority of structural and functional similarity to mineral composition of nature bones. 16 Our previous study demonstrated that one-dimensional HAp short nanowires had admirable osteogenic potential by nanostructure stimulation without any growth factors assistance. 17 Therefore, HAp short nanowires are suitable as osteoinduction layer for repairing subchondral bone in the biphasic structure.
Recently, Silk fibroin (SF)-based materials are emerging as an excellent matrix for cartilage tissue engineering. [18][19][20] The 3D printed SF-glycidyl-methacrylate hydrogel enhanced chondrogenic differentiation of encapsulated cells and the formation of new cartilage-like tissue. 21 However, many SF-based scaffolds were short of desired mechanical properties and unable to bear and buffer timely. Even after crosslinking, elastic modulus of most hydrogels is only about 0.01-0.2 MPa, [22][23][24][25][26] which is far from natural cartilage tissue (~4.1 MPa). 27,28 Chondroitin sulphate (CS), a main glycosaminoglycan in cartilage extracellular matrix (ECM), is extensively used for treating osteoarthritis and cartilage tissue engineering. 29 Besides intrinsic biological activities, physical cues, such as surface nanotopography and mechanical properties of materials, have synergistic effects on committed differentiation and tissue regeneration since the physical microenvironment for cell survival had tissue-specific topography and rigidity. 30,31 In this study, we prepared an SF-CS/HAp bilayer membrane following a biomimetic microstructure engineering design principle.
The physicochemical properties, releasing profile and cell compatibility were characterized. Subsequently, both in vitro and in vivo studies were implemented to assess the effectiveness of the proposed design principle.

| Preparation of SF-CS/HAp membrane
HAp short nanowires were synthesized according to our previous method. 17 Degumming process was in accordance with a previous study. 32 Degumming SF and CS were added into formic acid solution containing CaCl 2 to form SF-CS solution. HAp short nanowires dispersed in ethanol were poured into a compatible mould, and HAp film was formed after dry. The SF-CS solution was poured to HAp film. After the solvent evaporation, SF-CS/HAp composite was immersed into deionized water to remove CaCl 2 , followed by solidifying with ethanol.

| Physicochemical characterization of SF-CS/ HAp membrane
The morphology and structure characteristics were observed under an S-4800 scanning electron microscope (SEM) (Hitachi) and a JEM-2100 transmission electron microscope (TEM) (Jeol). To analyse the crystallinity, X-ray diffractograms (XRD) were performed on a Bruker D8 advance powder diffractometer equipped with a Cu Kα sealed tube. Chemical composition was assessed by a fourier transform infrared spectroscopy (Thermo Scientific). The tensile mechanical properties of the membrane (8 × 10 × 0.4 mm 3 ) were measured by a universal testing machine (Instron 3340). All samples were stretched at a speed of 0.437 mm/s. Nanoindentation measurements were performed using a three-sided pyramidal Berkovich diamond indenter with a nominal edge radius of 20 nm (faces 65.3° from vertical axis) attached to a fully calibrated nanoindenter (Nano Indenter G200, Agilent). The indentations were conducted with a continuous stiffness measurement (CSM) module and a speed of 10 nm/s. Topographic features and roughness of SF-CS surfaces were studied under an atomic force microscope (AFM; Dimension Icon, Bruker). The root mean square average (Ra) roughness was calculated from 0.5 × 0.5 μm 2 image areas.

| CS release from SF-CS/HAp membrane
SF-CS/HAp membrane was soaked into 2 mL phosphate-buffered saline (PBS, HyClone) in a centrifuge tube and then placed in a constant temperature oscillator (37°C, 100 rpm). At the desired time intervals (3,5,7,9,11,13,17 days), 2 mL PBS was extracted and equivalent fresh PBS was replenished to the tube. The absorbance of the collected PBS was measured at 260 nm wavelength using an ultraviolet spectrophotometry. The release profile was obtained by calculating the concentrations of CS according to the standard curve.

| Cell culture and cytocompatibility evaluation of SF-CS films
Bone marrow mesenchymal stem cells were harvested from healthy male SD rats (3-4 weeks) by direct panmyeloid adherence. The

| Osteogenic immunofluorescence staining
After cultured for 14 days, BMSCs on different sides of the membrane were collected by trypsinization and seeded into a new culture plate. Cells were then fixed with 4% paraformaldehyde and permeabilized using 0.1% Triton X-100 (Solarbio). After blocking with 10% normal goat serum, cells were incubated with an anti-osteocalcin (OCN) primary antibody (Abcam) and an anti-osteopontin (OPN) primary antibody (Abcam) at 4°C overnight. Cy3-conjugated goat anti-rabbit and Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody were used for OPN and OCN staining in the dark for 1 hour, respectively. Images were observed under a fluorescence microscope (OLYMPUS IX73).

| RNA isolation and quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)
Respectively, BMSCs were seeded onto SF-CS sides with chondroinductive medium (CM) and HAp sides with osteoinductive medium (OM) for 14 days. Total RNA of the cells was isolated using TRIzol reagent. RNA was reverse-transcribed to complementary DNA, and qRT-PCR was performed with LightCycler 96 Real-Time PCR System (Roche) using SYBR ® Premix Ex Taq™ II (Takara) to detect the gene levels of OCN, OPN, collagen type II alpha 1 (COL2A1), Aggrecan (ACAN) and sex-determining region Y box protein 9 (SOX-9). The relative transcript levels of the target gene expressions were normalized to β-actin and expressed as mean ± SD (n = 3).

| Alcian blue staining
Bone marrow mesenchymal stem cells were seeded onto SF sides containing 4%, 8%, 12% and 16% CS in CM for 14 days. Cells were washed with PBS thrice and fixed with 4% paraformaldehyde.
Subsequently, cells were stained with Alcian blue staining solution (Solarbio) and the nuclei were counterstained with redyeing solution. Images were obtained using a microscope (OLYMPUS BX53). Eight-week-old male SD rats were used in this study. A cylindrical osteochondral defect (with 1.5 mm diameter and 1 mm depth) was created with a dental drill at the trochlear grooves of the distal femurs. The rats were randomly divided into five groups, and different membranes were implanted into the defects: (a) negative control (NC); (b) SF membrane; (c) SF/HAp membrane; (d) SF-CS membrane;

| Animal osteochondral defect models
(e) SF-CS/HAp membrane. At 6 and 12 weeks postoperatively, the rats were sacrificed by excessive pentobarbital anaesthesia and fixed with 4% paraformaldehyde by cardiac perfusion. The distal femurs of the rats were harvested for the following experiments.

| Gross observation and micro-computational tomography (micro-CT) analysis
Images of gross specimen were captured using a Digital SLR cameras (Canon). The effect of cartilage rehabilitation was evaluated by three blinded independent researchers according to ICRS

| Histological analysis
H&E staining, toluidine blue staining, Saf-O staining and immunohistochemical staining with polyclonal rabbit anti-Col-II antibody and anti-Col-I antibody (Abcam) were performed to appraise cartilage regeneration in line with the manufacturer's protocols. The specimens were observed under a BX53 microscope. The cartilage regeneration at week 6 and 12 was assessed by histological scoring system for evaluation of cartilage repair. 33 The immunohistochemical staining intensity for Col-II and Col-I was calculated by Image-Pro Plus 6.0 software.

| Statistical analysis
Data were presented as mean ± standard deviation (SD). Differences between more than two experimental groups and NC group were analysed by one-way ANOVA followed by Tukey's HSD comparison test, and variance between two groups was compared by twoway t test with GraphPad Prism software (version 6, by MacKiev Software). P < .05 was considered statistically significant.

| Characterization of SF-CS/HAp membrane
of the bilayer structure had irregular nanotopography ( Figure 1F), and HAp side maintained the original morphology of short nanowires as mentioned above ( Figure 1H). In addition, the inseparable combination of SF-CS and HAp realized a seamless interface between the organic and inorganic phases, which was based on the permeation of with an initial burst release (~20%) at the first 3 days from the surface, and the releasing reached a balance at day 11 ( Figure S4).

| Cell viability
Pure SF, 4% and 8% CS-loaded SF did not affect BMSCs proliferation after incubation for 48 hours. Although 12% and 16% CS inhibited proliferation (P < .01), BMSCs still kept relatively high viability ( Figure 3A). To observe cell status visually, BMSCs were incubated for 48 hours and stained with a Live/Dead kit (red indicates dead cells; green indicates live cells). Cells seeded on SF-8% CS membrane exhibited high viability with a majority of live cells and rare dead cells, which was comparable with those cultured on TCP ( Figure 3B).

| Osteogenic and chondrogenic differentiation capacity assessment
To evaluate osteogenic capacity, BMSCs were separately inoculated onto TCP and two sides of SF-CS/HAp membrane within OM for

| In vivo promotion effect on osteochondral defect repair
Gross view, micro-CT and histological analysis were used to confirm in vivo regenerative potential of SF-CS/HAp membrane on osteochondral defect model. Figure

| D ISCUSS I ON
Concerns are growing about biomimetic restoration, and difunctional structure with concomitant chondrogenic/osteogenic potential draws increasing attention. In recent years, the concept of difunctional biomimetic repairing has been penetrated into the osteochondral tissue engineering scaffolds. [34][35][36] In this study, bilayer SF-CS/HAp membrane imitating the structure and function of osteochondral unit was developed to obtain osteochondral repairing, and the repairing ability was assessed in vitro and in vivo. Mechanical strength is one of the most critical factors for assessing the properties of cartilage repairing biomaterials, and sufficient mechanical strength is essential for bearing heavy loads and buffering stress. 2 However, it is challenging to meet both bio-properties and desired mechanical performance for biomimetic scaffolds in osteochondral repairing. In our study, tensile and compressive modulus (~5.26 and ~6.7 MPa) of SF-CS/HAp membrane were in a reasonable range for the cartilage regeneration, and comparable to that of the natural cartilage tissue. 27,28 Ethanol-induced β-sheet structure was responsible for the favourable mechanical performances. 42 In addition, SF could serve as a sustained release matrix for growth factors or protein drugs and preserve their potency successfully. 43 The sustained release pattern of CS from SF-CS/HAp membrane may be relevant to slowly biodegradable property of SF. 32 As a naturally occurring protein polymer, SF possesses excellent biocompatibility, which is conducive to cell survival. 18  The bio-functionalization of developed biomaterials was conducive to further enhance its bioactivity. It is well known that surface nanostructures or decoration of bioactive motifs could regulate cells behaviours, and greatly improve the potential of cell differentiation, which is the key factor to the success of the implanted scaffolds. 17,44,45 For example, a noncanonical Wnt5a motif could functionalize biomaterials in enhancing the osteogenesis and associated skeleton formation. 44 After osteogenic induction, cells on HAp sides presented superior osteogenic differentiation potential as compared to SF sides and TCP, which may be due to the beneficial physical stimulation from the nanotextures of HAp surface. 30,31 In addition, the stiffness of HAp may also contribute to osteogenic differentiation. 46  Potentially, the best cartilage regeneration capability of SF-CS/HAp structure is not only ascribed to the chondrogenic induction from SF-CS layer but also to the osteogenic promotion by HAp nanowire layer. Therefore, the bionic structure could serve dual functions for osteochondral regeneration.
In this study, we successfully developed a nanotextured SF-CS/ exist with regard to this study.

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.