DBM frameworks (Daqing Bio Co. Ltd., Chongqing, China) were cut into 7-mm-long, 5-mm-wide, and 2.5-mm-thick cuboid constructs. The PCL framework was fabricated using a 3D-printer. The properties of the PCL framework were designed using CAD and Mimics 17.0 software, and the frameworks were printed using a 3D layer-by-layer fused deposition modeling (FDM) printer (FoChif Tech HTS, China). In brief, PCL pellets (Mw 80,000, Sigma, United States) were melted (at 120°C) in a printing chamber and then printed with a lay down pattern of 0°/45°/90°/135° (top layer) or 0°/90° (bottom layer) along the z-axis. Thus, a double-layered PCL framework model was produced and then subsequently cut into cuboids with length, width, and height of 7, 5, and 2.5 mm, respectively. All frameworks were sterilized using ethylene oxide before use. The mechanical analysis of PCL and DBM framework was carried out using a mechanical testing machine (Instron-5542, Canton, MA, United States). All samples (n = 5 per group) were processed into a cuboid shape, and a constant compressive strain rate of 0.5 mm/min was applied until 80% of the maximum deformation. The stress and strain curves were obtained from the first 40%. The Young’s modulus was calculated from the stress and strain curves. The endotoxin content in the leach solutions of the frameworks were analyzed using a chromogenic endpoint Tachypleus amebocyte lysate (TAL) assay kit (Xiamen Houshiji, China), following the kit instructions as previously described (Wei et al., 2015).

A total of three 6-month-old goats (Shanghai Jiagan Biological Technology Co., Shanghai, China) were used in this study. All protocols of animal study were approved by the Animal Care and Experiment Committee of Shanghai Jiao Tong University School of Medicine.

Cartilage sheets were prepared as previously reported (Li et al., 2017). Furthermore, two- or three-passage goat chondrocytes were harvested, suspended, and then seeded in six-well cell culture plates at a density of 1.5 × 10⁷ cells/well. The chondrocytes were then cultured in a chondrogenic medium, containing 100 ng/ml IGF-I (R&D Systems Inc. Minneapolis, MN, United States), 10 ng/ml TGF-b1 (R&D Systems Inc. Minneapolis, MN, United States), 40 ng/ml dexamethasone (Sigma-Aldrich, St. Louis, MO, United States), 1% insulin–transferrin–selenium–linoleic acid (ITS, ScienCell, CA, United States), and 1% antibiotic–antimycotic (Gibco BRL) in DMEM for 5 days. The cartilage sheets were then minced into a gelatinous mass and collected in a syringe before being seeded in their respective frameworks to form constructs. The constructs were incubated for 2 h and then transferred into a 6-well plate containing the culture medium. After 3 days, the constructs were subcutaneously implanted in autologous goats. The cellular viability of the cartilage sheets and minced cartilage sheet (ECG) was evaluated using the Live/Dead Cell Viability Assay (Invitrogen, Carlsbad, CA, United States), following the manufacturer’s instructions, and examined by confocal microscopy (Nikon, Japan).

After in vitro culture for 3 days, both constructs (PCL-ECG and DBM-ECG, n = 15 constructs per group in each goat) were subcutaneously implanted in autologous goats. As control groups, frameworks without ECG (DBM and PCL framework, n = 15 frameworks per group in each goat) and ECG without a framework were also implanted and injected into the goats, respectively. During surgery, each goat was anesthetized and endotracheally intubated. The constructs and frameworks were implanted in subcutaneous pockets made in the abdominal area. ECG without framework was injected using a syringe. The animals were allowed to recover from anesthesia after closure of the incisions. Samples were harvested at 1, 4, and 8 weeks postimplantation for gross, histological, immunohistochemical, and quantitative evaluation.

After 1 and 4 weeks of implantation, samples from all groups (n = 5 samples per group in each goat) were harvested for analysis of the inflammatory response. After gross observation, samples were fixed in 4% paraformaldehyde for 48 h, then embedded in paraffin, and sectioned into 5-mm-thick slices. Slices were stained with hematoxylin and eosin (HE). For immunohistochemical analysis, CD68 was detected using mouse anti-CD68 monoclonal antibody (ab955, 1:200, Abcam), followed by goat anti-mouse IgG H&L (HRP) (ab205719, 1:2000, Abcam). Apoptotic cells were detected by terminal deoxynucleotidyl transferase biotin-dUTP nick end labeling (TUNEL) using a TUNEL kit (Roche, Indianapolis, IN, United States), following the manufacturer’s instructions. The quantification of CD68 and TUNEL position area (%) was performed using ImageJ and IHC Profiler software (n = 5 per group).

For fluorescent staining, the cells were permeabilized with 0.1% Triton X-100 (Sigma-Aldrich) for 5 min at room temperature, washed with PBS, and then stained with DAPI and FITC–phalloidin. FITC and phalloidin (Sigma–Aldrich) were diluted in PBS in a 1:200 ratio and incubated on the samples away from light for 30 min. After incubation, the samples were washed with PBS three times. Cell nuclei were stained with DAPI for 8 min, after which the samples were washed with PBS three times. Imaging of RAW 264.7 cells was performed using a fluorescence confocal microscope (Nikon, Japan).

Cell culture supernatant (1 ml) was collected after 24, 48, and 72 h of incubation and used for enzyme-linked immunosorbent assay (ELISA). Mouse IL-6 ELISA Kit, Mouse TNF-α ELISA Kit, and Mouse Cox-2 ELISA Kit (all ELISA kits; Invitrogen, Thermo Fisher Scientific, Waltham, MA, United States) were used according to the manufacturer’s instructions.

Gene expression of inflammatory cytokines was investigated using the real-time polymerase chain reaction (RT-PCR). The expression levels of IL-6, TNF-α, and Cox-2 genes were analyzed. Each group of RAW 264.7 cells was collected, and the total RNA was extracted using the TRIzol reagent (Invitrogen), after which the total RNA was reverse transcribed using Moloney murine leukemia virus reverse transcriptase (Invitrogen). qPCR was performed using a Fast Synergy Brands Green Master Kit and a Light Cycler 480 system (Roche), following the manufacturer’s instructions. The forward and reverse primer sequences are listed in Table 1. The results were analyzed using the comparative threshold cycle method and normalized to the endogenous reference gene β-actin.

After 8 weeks of culturing in vivo, samples of the PCL-ECG, DBM-ECG, and ECG groups (n = 5 samples per group in each goat) were carefully extracted from the goats. After gross observation and measurement, part of each sample (the rest of the sample was used for subsequent biochemical analysis) was fixed in 4% paraformaldehyde, embedded in paraffin, sectioned to give 5-mm thicknesses, and then mounted on glass slides for histological and immunohistochemical analyses. The slices were stained with H&E and safranin-O to evaluate the histological structure of the engineered cartilage (EC), and for the immunohistochemical analysis, expression of type II collagen (COL II) was evaluated to determine ECM deposition of the ECs using rabbit anti-collagen II polyclonal antibody (ab34712, 1:100, Abcam) with goat anti-rabbit IgG H&L (HRP) (ab205718, 1:2000, Abcam) as a secondary antibody. Quantification of the regenerated cartilage area (%) was performed using ImageJ and IHC Profiler software (n = 5 per group).

Quantitative analysis was performed as previously described (Jia et al., 2020). In brief, an electronic balance was used to measure the weight of all samples (n = 5 per group). The volume of each sample was measured using the water displacement method (n = 5 per group). The total glycosaminoglycan (GAG) content and the total collagen content of the samples (n = 5 per group) was quantified using the alcian blue method and hydroxyproline assay, respectively.

Statistical analyses were performed using SPSS 23 (IBM, United States). Student’s t-test was performed to compare the mechanical properties and ECG loading efficiency of the frameworks. One-way ANOVA was performed to compare the results of the cytotoxicity evaluation of the frameworks and immunohistochemistry results. Two-way ANOVA was performed to test the interaction between two independent variables (time and material type for the quantitative analysis results, and time and leach solution type for the quantitative RT-PCR and ELISA results). Tukey’s honestly significant difference (HSD) post hoc tests were performed after ANOVA. Data are presented as the mean ± standard deviation (SD). The value of ∗p < 0.05 was considered statistically significant.

Fabrication and characterization of the PCL framework was the first step of PCL-ECG construct preparation. As shown in Figure 1, a double-layered PCL framework structure was 3D-printed and assembled. The top layer with large pores was suitable for ECG loading, while the bottom layer with small pores was designed to prevent ECG loss (Figures 1A,B). Both SEM and pore size analysis revealed that the PCL framework had a uniform pore structure, while the DBM framework presented a clearly heterogeneous structure with varied pore sizes (Figures 1D,E), indicating the relative homogeneity and controllability of the PCL framework in terms of ECG distribution, mechanical properties, degradation rate, and shape maintenance. The mechanical analysis revealed that the mechanical strength of the PCL framework was significantly higher than that of the DBM framework (Figure 1F). The PCL framework had a mechanical strength close to that of native ear cartilage (Zhou et al., 2018), which may help it to maintain the original shape of the regenerated cartilage.

The biocompatibility was then evaluated by loading ECG into the frameworks. Similar to our previous reports, cartilage sheets cultured in vitro for 5 days presented soft fragile membranes that could be easily collected, minced into gel form (Figures 2A,B) while maintaining good cellular viability (Figures 2C,D), and then loaded into the frameworks (Figure 2E). SEM showed that ECG adhered well to the frameworks and completely covered the frameworks after 3 days of in vitro culture owing to the abundant ECM production (Figures 2F,G). The quantification analysis revealed that the ECG loading efficiencies in the two frameworks were both higher than 90% with no statistical difference (Figure 2H), indicating good cytocompatibility for both frameworks. Cytotoxicity evaluation showed that cartilage sheets survived well in the leach solutions of both frameworks with visible cell proliferation over time (Figures 3A,B). Notably, few dead cells were observed in the group treated with PCL leach solution (no significant difference with the DMEM control group, Supplementary Figure S1), while some dead cells were found in the group treated with DBM leach solution (Figure 3C). This indicates the higher cytotoxicity of the DBM framework compared with the PCL framework, which might be related to higher endotoxin residue in the DBM framework (Supplementary Figure S2).

In vivo inflammatory responses were evaluated to predict the feasibility of cartilage regeneration using the PCL-ECG constructs. The observations after 1 week show that all samples were easily extracted from the implantation sites without obvious adhesion to peripheral tissues (Figure 4A), suggesting relatively low inflammatory responses in all groups. H&E and CD68 immunohistochemical staining showed that the inflammatory responses for both the PCL-ECG and DBM-ECG groups were much stronger than that for the ECG group, indicating that both the PCL and DBM frameworks showed some immunogenicity. This was further supported by the inflammatory responses triggered by the implantation of frameworks without ECG (Figures 4B–D). TUNEL immunohistochemical staining also confirmed that more apoptotic cells caused by inflammatory response were found for the groups containing frameworks than for the ECG group (Figure 4E). It was worth noting that the PCL framework triggered a milder inflammatory response with less cell apoptosis than the DBM framework (Figures 4D,E), which was further confirmed by a semiquantitative analysis (Figures 4F,G), indicating that the PCL framework had lower immunogenicity than the DBM framework. As anticipated, the intensity of the inflammatory responses at 4 weeks was lower than those for the samples after 1 week, with less cell apoptosis in all groups (Figures 5A–E). Semiquantitative analysis further revealed that both the ECG and PCL-ECG groups presented minimal inflammatory infiltration and cell apoptosis (with no significant difference), while the DBM framework still exhibited higher levels of inflammatory response and cell apoptosis (Figures 5F,G). This indicates that the PCL framework is more suitable for supporting cartilage regeneration of ECG than the DBM framework in terms of immunogenicity.

An in vitro macrophage M1 polarization experiment was conducted to further evaluate the framework immunogenicity. As shown in Figure 6, RAW 264.7 cells cultured in DMEM presented relatively uniform morphology with small round cortical actin rings (Figure 6A), while the cells treated with LPS exhibited typical M1 polarization morphology with large dendritic cortical actin rings (Figure 6B). Cells cultured in the leach solution of the PCL framework showed minor morphology changes essentially maintaining small round cortical actin rings (Figure 6C), while cells cultured in the leach solution of the DBM framework presented a discernible morphology change, exhibiting relatively large and irregular cortical actin rings (Figure 6D). The results of qPCR and ELISA were consistent with the cell morphology changes, further confirming that the expressions of M1 polarization related cytokines (IL-6, COX-2, and TNF-α) in the PCL group were significantly lower than those for the DBM group in terms of both gene and protein levels (Figure 7). These results indicate that the leach solution of the PCL framework triggered milder M1 polarization of macrophages than the leach solution of the DBM framework, suggesting that the PCL framework showed lower immunogenicity than the DBM framework.

The in vivo cartilage regeneration performance was analyzed to evaluate the clinical translational potential of the PCL-ECG strategy. Gross observation showed that the regenerated cartilage in all groups gradually matured as evidenced by a reddish appearance at 1 week to an ivory appearance at 8 weeks (Figures 4A, 5A, 8A). The samples in the PCL-ECG group essentially maintained their original shape and size, showing relatively regular cuboids, while the samples in the DBM-ECG group showed slight deformation with an irregular cuboid shape (Figure 8A). As anticipated, the samples in the ECG group showed an irregular shape due to lack of a supporting framework (Figure 8A). Histological analysis revealed that the samples in all groups formed mature cartilage-like tissue with typical lacuna structures and abundant cartilage-specific ECM deposition evidenced by strong positive staining of safranin-O and collagen II (Figures 8B–E). The quantitative analysis showed that the GAG and total collagen content of the regenerated cartilage in all groups showed a gradually increasing trend, indicating the gradual maturation of neo-cartilage, which was further supported by the gradually increasing wet weights and volumes observed for both the PCL-ECG and DBM-ECG groups. Notably, the wet weight and volume of the ECG group decreased over time, which might be related to stress induced absorption due to lack of a supporting framework (Supplementary Figure S3). In addition, it is worth noting that all of the quantitative data for the PCL-ECG group were higher than those for the DBM-ECG group (Figure 9), indicating relatively higher cartilage yield in the PCL-ECG group. These results suggest excellent clinical translation potential for the PCL-ECG cartilage regeneration strategy.