An Avascular Niche Created by Axitinib‐Loaded PCL/Collagen Nanofibrous Membrane Stabilized Subcutaneous Chondrogenesis of Mesenchymal Stromal Cells

Abstract Engineered cartilage derived from mesenchymal stromal cells (MSCs) always fails to maintain the cartilaginous phenotype in the subcutaneous environment due to the ossification tendency. Vascular invasion is a prerequisite for endochondral ossification during the development of long bone. As an oral antitumor medicine, Inlyta (axitinib) possesses pronounced antiangiogenic activity, owing to the inactivation of the vascular endothelial growth factor (VEGF) signaling pathway. In this study, axitinib‐loaded poly(ε‐caprolactone) (PCL)/collagen nanofibrous membranes are fabricated by electrospinning for the first time. Rabbit‐derived MSCs‐engineered cartilage is encapsulated in the axitinib‐loaded nanofibrous membrane and subcutaneously implanted into nude mice. The sustained and localized release of axitinib successfully inhibits vascular invasion, stabilizes cartilaginous phenotype, and helps cartilage maturation. RNA sequence further reveals that axitinib creates an avascular, hypoxic, and low immune response niche. Timp1 is remarkably upregulated in this niche, which probably plays a functional role in inhibiting the activity of matrix metalloproteinases and stabilizing the engineered cartilage. This study provides a novel strategy for stable subcutaneous chondrogenesis of mesenchymal stromal cells, which is also suitable for other medical applications, such as arthritis treatment, local treatment of tumors, and regeneration of other avascular tissues (cornea and tendon).

Fabrication of axitinib-loaded PCL/collagen nanofibrous membranes: PCL and collagen were dissolved in HFIP at a mass ratio of 60/40 and a concentration of 12% (wt/v), and then stirred at room temperature for 24 h. Then, 1 h before electrospinning, axitinib was added to the solution at mass ratios of 0%, 1%, 3%, and 6% ( weight of axitinib total weight of PCL and collagen ). The electrospinning conditions were as follows: injection rate was 1.5 mL/h, 30% to 50% wet, the voltage was 10 kV, and distance between the syringe needle and the grounded rotating plate (300 rpm) was 11 cm. Nanofibrous membranes were vacuum freeze-dried for 48 h and then preserved at -40 °C before use.

Characterization of PCL/collagen nanofibrous membranes:
The surface morphologies of different membranes were observed using a scanning electron microscope (SEM; TM 3030 Plus, Hitachi). More than 100 random nanofibers were measured by Nano Measurer software to determine the average fiber diameter. The nanofibrous structures and distribution of axitinib were observed using a transmission electron microscopy (TEM; JEM2100, Japan).
The hydrophilicity of the different membranes was evaluated using a hydrophilic angle tester (Attension Theta, Biolin Scientific AB, Sweden). Deionized water (5 µL) was dropped carefully onto the surface of different membranes. The contact angle was measured at random locations.
Attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR) of axitinib powder and nanofibrous membranes were performed using a Nicolet-670 FTIR spectrometer (Thermo Scientific, USA). All spectra were recorded at the 1000-4000 cm -1 wavelength range.
The encapsulation efficiency of axitinib was measured as follows: [1] a known mass of the membrane was dissolved in 1 mL HFIP. 100 uL of the solution was then added to 2 mL of methanol and centrifuged to remove the precipitated PCL and collagen. 20 uL of the diluted solution was detected using a high-performance liquid chromatography (HPLC) instrument (1260 Infinity II Prime, Agilent, USA) at λ = 330 nm. The amount of axitinib was obtained from the standard curve of axitinib. The encapsulation efficiency was calculated using the following equation: Encapsulation efficiency= weight of axitinib in the sample theoretical weight of axitinib loading in the sample

100%
The release profile of axitinib was determined by soaking the membranes in PBS in vitro.
The membranes were cut into squares, accurately weighed, and incubated in 5 mL PBS at 37 °C while shaking mildly. Every 3 to 6 days, all solutions were collected for HPLC detection to determine the exact amount of axitinib released, which was then replaced with an extra 5 mL PBS. The cumulative release rate was calculated based on the actual weight of the axitinib encapsulated in the membranes.
Cell isolation and culture: Human umbilical vein endothelial cells (HUVECs) were purchased from Cyagen Biosciences (USA). The culture medium for HUVECs was composed of DMEM with high glucose, 10% FBS, and antibiotic-antimycotic (1X).
The medium was replaced every 2 to 3 days until the primary BMSCs reached a confluence of approximately 80%. Then, the BMSCs were collected by 0.25% trypsin and subcultured at a density of 1 × 10 4 cells/cm 2 . BMSCs at passage 2 were used for further experiments.  Subsequently, membranes were placed at the bottom of 24-well culture plates and fixed with steel rings. The HUVECs were seeded on the membranes at a density of 1 × 10 4 cells/well. 1 or 5 days after cell seeding, HUVECs were washed thrice with PBS, fixed with 2.5% glutaraldehyde, and stained with DAPI. The cell density of HUVECs was directly observed using a fluorescent inverted microscope (DMI300B, Leica, Germany).
The Edu assay was performed 1 day after cell seeding to evaluate the cell proliferation potential. Briefly, 30 µM Edu was added to the cell culture medium and incubated with HUVECs for 2 h at 37 °C. HUVECs were fixed, permeabilized, and stained with Azide-594 and then counterstained with DAPI. The percentage of Edu-positive cells was analyzed using a fluorescent inverted microscope. More than 6 high-power fields were countered to determine the proliferation potential of HUVECs on different membranes. The morphology of HUVECs was also observed via SEM. Briefly, 5 days after seeding, HUVECs seeded on membranes were fixed with 2.5% glutaraldehyde, dehydrated with graded alcohols, and dried naturally. Then, samples were coated with gold and observed under an electron microscope.

Biocompatibility of axitinib-loaded PCL/collagen membranes with BMSCs and
chondrocytes: Membranes were prepared as described above. BMSCs or chondrocytes were then seeded onto membranes at a density of 1 × 10 4 or 2 × 10 4 cells/well, respectively.
To directly analyze the morphological features of BMSCs or chondrocytes on membranes, the cytoskeleton was stained with phalloidin (PHDH1, Cytoskeleton, USA) and counterstained with DAPI 1 day or 7 days after cell seeding. To evaluate the proliferation activity, the CCK-8 test was performed 1, 4, and 7 days after cell seeding.

Construction and evaluation of the vitro-BEC:
The PLGA scaffolds were tailored and compressed into cylindrical shapes (6 mm diameter and 2 mm depth) and then disinfected with 75% ethanol solution for 1 h and washed thrice with PBS. Thereafter, 2 nd passage BMSCs were suspended in enrichment medium, as previously described, [3] to a final concentration of 6 x 10 7 cells/mL, and 65 µL of the cell suspension was evenly seeded into each scaffold, followed by a 5 h incubation. After two days of cultivation in the enrichment medium, the constructs were cultured in the chondrogenic medium composed of DMEM with high glucose, 10 ng/mL TGFβ1, 100 ng/mL IGF1, ITS (1X), 25 µg/mL ascorbic acid, 40 ng/mL dexamethasone, and antibiotic-antimycotic (1X) for 4 weeks to prepare the BEC.
Finally, the chondrogenesis in the BEC was evaluated through histological examination.

Encapsulation of the vitro-BEC and implantation in vivo:
The BEC was encapsulated in PCL/collagen nanofibrous membranes with a small amount of tissue adhesive (Histoacryl, B. Braun Melsungen AG, Germany). First, BEC was placed onto the center of the membrane.
Second, the four corners of the membrane were folded to the upper surface of the BEC.
Chondrogenesis assessment of the engineered cartilage: The glycosaminoglycan (GAG) and Collagen II contents were chosen as the indices of chondrogenesis. After 24 weeks of subcutaneous implantation, samples were harvested from each group. The GAG content was quantified using dimethyl methylene blue chloride (DMMB, Sigma, USA), [2] and the Collagen II content was quantified using an ELISA kit (ml1001622-C, Enzyme-linked Biotechnology, Shanghai, China), according to the manufacturer's instructions. Vitro-BEC was used as a basal control, while native auricular cartilage was considered as a positive control.
Evaluation of self-stability of engineered cartilage: mRNA expression levels of engineered cartilage were analyzed in the 3%-Axitinib group (in vivo 24 weeks, n=3) and compared with those of vitro-BEC (n=3). Rabbit Sox9, Col2a1, Aggrecan, Chm-I, Runx2, and Vegfa were chosen to estimate the chondrogenesis and ossification tendency, with beta-actin (Actb) was used as an internal control. Primer sequences were shown in Table S2, supporting information.
Further, to judge whether the engineered cartilage was able to remain stable alone, we stripped the membrane surrounding the engineered cartilage in the 3%-Axitinib group (in vivo 20 weeks, n=6).The residual axitinib in the membrane was measured by HPLC and the nude engineered cartilage was subcutaneously implanted into a new nude mouse. After 2 or 4 weeks of implantation, samples were harvested and evaluated by the overall view and pathological staining.

RNA sequence and qPCR analysis of murine tissues surrounding engineered cartilage:
After 12 weeks of subcutaneous implantation, samples from the 0%-Axitinib group and 3%-Axitinib group were harvested, with the murine tissues entangled. Total RNA was extracted using Trizol and evaluated by agarose gel electrophoresis and Agilent 2100 Bioanalyzer (USA). After the RNA samples were qualified, mRNAs were enriched using magnetic beads with Oligo (dT). Isolated mRNAs were then fragmented into short pieces, reverse transcribed into cDNA, purified, terminally repaired, ligated to the adapter, and amplified. Raw mRNA sequence data were obtained using an Illumina HiSeq TM 2000 machine (USA) and filtered using Fastp software.
First, the trimmed reads were mapped to the mouse genome (mm10, genome.UCSC.edu) using the STAR software to identify the transcriptional information in the niche of BEC. Then, the DESeq2 "counts" function of the R package was used to normalize gene expression levels based on the number of total reads of each sample. Differentially expressed genes between the 0%-Axitinib group and 3%-Axitinib group were identified as meeting the criteria of fold change ≥ 1.5 and P-value＜0.05. Lastly, we used Metascape (http://metascape.org) for gene ontology analysis.
For qPCR analysis, murine mRNAs were reverse transcribed into cDNA. Primers for target genes (Pecam1, Cdh5, Timp1, Mmp9 and Mmp13) were designed, shown in Table S3 Statistical analysis: All quantitative data were obtained from no less than three samples and presented in terms of their mean ± standard deviation. The Anderson-Darling test was first performed to evaluate if the raw data fit the normal distribution curve. If this fit did not exist, logarithmic transformation was done on this data. A Student's t-test was used to determine statistical differences between two groups, while a one-way analysis of variance (ANOVA), followed by a Tukey's test, was conducted for multiple group comparisons. Significant differences were assumed at p-values below 0.05 (*p < 0.05; **p < 0.01; ***p < 0.001). All statistical analyses were performed using the software SPSS Statistics 24.0 (IBM, USA).        Figure S10. Immunohistochemistry staining of Mmp9, Mmp13, and Timp1 in 0%-Axitinib and 3%-Axitinib groups at 12 weeks post-implantation. Significant positive staining of Mmp9 or Mmp13 was observed in both 0%-Axitinib and 3%-Axitinib groups (a-d). However, the expression of Timp1 in the 0%-Axitinib group (e) was much lower than that in the 3%-Axitinib group (f). Black arrows represented murine tissues surrounding engineered cartilage. Scale bar = 50 μm.  At the scheduled time, 6 samples in each group were evaluated by pathological staining. a) indicated that typical positive staining of safranin O was observed; b) indicated that typical positive staining of fast green was observed; c) indicated that typical bony structures consisting of bone trabecula were observed.