Chondroitin Sulphate and Gas Foaming Concurrently Improve Articular Cartilage Regeneration in Electrospun Poly(L-lactide-co-ε-caprolactone)/Silk Fibroin Based Scaffolds

Degenerated cartilage tissues remain a burgeoning issue to be tackled, while bioactive engineering products available for optimal cartilage regeneration are scarce. In the present study, two-dimensional (2DS) poly(L-lactide-co-ε-caprolactone)/silk broin (PLCL/SF)-based scaffolds were fabricated by conjugate electrospinning method, and then cross-linked with chondroitin sulfate (CS) to further enhance their mechanical and biological performance. Afterwards, three-dimensional PLCL/SF scaffolds (3DS) and CS-crosslinked three-dimensional scaffolds (3DCSS) with tailored size were successfully fabricated by in situ gas foaming in a conned mold and subsequently freeze-dried. Gas-foamed scaffolds exhibited high porosity, rapid water absorption, and stable mechanical properties. While all of the scaffolds exhibited excellent cytocompatibility in vitro; 3DCSS showed better cell seeding eciency and chondro-protective effect as compared to the other scaffolds. Histological analysis of chondrocytes-seeded constructs after cultivation for up to 6 weeks in vitro also conrmed that 3DCSS scaffolds supported the formation of cartilage-like tissues along with the more secretion of cartilage-specic extracellular matrix than that of the other groups. The reparative potential of 3DCSS was further evaluated in an articular cartilage defect model in rabbits, which exhibited a well-integrated boundary and attenuated inammation demonstrating less expression of pro-inammatory cytokines, such as interleukin (IL)-1β and tumor necrosis factor (TNF)-α. Taken together, the engineered biomimetic 3DCSS may provide a well-suited therapeutic option for cartilage tissue regeneration applications.


Introduction
Articular cartilage injuries, resulting from the acute or repetitive trauma, osteoarthritis and various other joint disorders, pose a substantial burden globally, which may cause permanent disability and discomfort [1]. Cartilage has a limited intrinsic healing capacity due to the lack of vascularization and innervation network as well as its isolation from the peripheral circulation. Therefore, cartilage injuries remain one of the most problematic diseases for orthopedic surgeons [2]. A variety of clinical approaches, such as drilling, mosaicplasty, microfracture, and autologous chondrocyte implantation (ACI), are proposed for restoring normal joint congruity and minimize further joint degeneration [3,4]. However, none of these techniques has achieved success for the functional cartilage repair owing to the prior surgical procedures, less numbers of donors as well as the low elasticity and toughness of neo-cartilage tissues [5,6]. To overcome the clinical shortages of synthetic cartilage substitutes, cartilage tissue engineering (CTE), has been put forwarded [7].
Since scaffolds play a signi cant role in CTE, it is essential to engineer an ideal platform exhibiting characteristics, such as biocompatibility, biodegradability, and su cient mechanical strength to support cell differentiation and matrix production [8]. Numerous strategies have been devised to afford CTE scaffolds, such as phase separation, 3D printing, and electrospinning; the latter has gathered signi cant attention of the research community owing to its applicability to a myriad of materials with high nano ber production e ciency [9]. Nonetheless, the morphology of the membranes assembled from normal electrospinning is generally comprised of randomly dense-packed ber layers with only super cial pores, preventing cellular in ltration necessary to form 3D tissues [10]. Conjugate electrospinning technology, which utilizes double metal spinnerets with opposing charges and forms nanoyarn scaffolds with a highly organized interior aligned structure, has been extensively researched to fabricate nanoyarn scaffold for annulus brosus or vascular tissue engineering [11]. Meanwhile, the simultaneous operation of two nozzles promotes the e ciency of the fabrication process. Although membranes assembled from conjugate electrospinning have achieved some improvements in the topological arrangement and porosity as compared to those fabricated by using conventional electrospinning, the ber organization is still tightly-packed limiting cellular in ltration [12]. Thus, there is a dire need to devise strategies to convert 2D nano brous membranes into 3D-like porous structures.
An array of post-treatment methods, such as gas foaming, self-assembly of short bers, and electrospraying have been proposed for transforming 2D membranes into 3D scaffolds [13,14]. Amongst, gas foaming, which either utilizes gas bubbles generated in situ via a chemical reaction (e.g., the decomposition of sodium borohydride, NaBH 4 ) or the addition of an inert gas, requires uncomplicated equipment, and can be used in almost all experimental settings, rendering it as an e cient method for fabricating 3D nano ber matrices with layered structures [15,16]. Although gas-foaming has been widely exploited for fabricating porous scaffolds, the selection of the synthetic material in most of the cases, including the use of hydrophobic polymers, such as nylon, poly(vinylene di uoride) (PVDF), and polycaprolactone (PCL) renders the scaffolds lacking cell recognition cues [17,18]. These scaffolds require further modi cation by plasma, or coating with extracellular matrix (ECM)-derived proteins to improve their hydrophilicity and cytocompatibility [19,20]. Yet in some seminal reports, natural and synthetic materials were combined to prepare hybrid gas-foamed scaffolds for wound healing or nerve regeneration, but how to improve their biological functionality, such as the induction of cell differentiation and promotion of speci c cellular matrix secretion is still a challenge [21,22]. It is known that the cartilage defect ampli es the in ammation in the joints, leading to an overproduction of proin ammatory mediators [23]. The establishment of a microenvironment that may promotes cartilage proliferation, maintains the cell phenotype as well as attenuates in ammation is therefore highly bene cial for cartilage regeneration. Furthermore, the control of the thickness of gas-foamed scaffold has not been paid su cient attention; it is di cult to precisely control the size of the scaffolds merely by varying the concentration of the foaming medium or the processing time.
In this study, we designed chondroitin sulfate (CS)-crosslinked 3D poly(L-lactide-co-e-caprolactone)/silk broin (PLCL/SF) scaffolds (3DCSS) with accurate thickness and evaluated their potential for cartilage regeneration. SF enhances the hydrophilicity of the hybrid scaffold and can sustain cell proliferation, while also possesses a low in ammatory potential [24]. CS is the physiological component of the cartilage, possessing numerous useful merits, including anti-in ammatory activity, water and nutrients absorption, chondrogenic potential at cellular level that helps restore structure and function of the articular cartilage [25]. Brie y, 2D PLCL/SF nano brous scaffolds (2DS) were afforded by conjugate electrospinning, followed by crosslinking with the CS by using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) and N-hydroxysuccinimide (NHS). Then, scaffolds were loaded into a manually fabricated mold and immersed into gas foaming solution (sodium borohydride, NaBH 4 ) to obtain porous 3DCSS with tailored size. We hypothesize that 3DCSS scaffolds feature precise thickness, better biomechanical and biological properties, which concurrently encourage cellular in ltration, chondri cation, cartilage-speci c ECM matrix secretion, and in ammation resolution. After a thorough characterization, including the composition, morphology, mechanical properties, cytocompatibility, and in ammatory response, the scaffolds were investigated in terms of the chondro-inductivity for up to 6 weeks in an in vitro culture as well as in an in vivo articular cartilage defect model in rabbits (Fig. 1

Fabrication of two-dimensional nano ber mats
A purpose-built electrospinning device (SS-3556H, Ucalery, Beijing, China) with a double-nozzle conjugated electrospinning system was utilized to fabricate two-dimensional PLCL/SF scaffolds (2DS) as described previously [26]. PLCL and SF were dissolved in HFIP in a mass ratio of 8:2 at a total concentration of 10 % (w/v). The mixed solution was loaded into two oppositely positioned blunt-ended needle 10-mL syringes pumped at a ow rate of 1.2 mL/h. A high-voltage with positive (+ 12 kV) and negative (-12 kV) static current was introduced on the needles of each side. During electrospinning process, the nano bers carrying opposite charges were intertwined in the air, and a rotating circular drum (500 rpm) was placed between two jets to collect the arranged 2DS.

Fabrication of 3D PLCL/SF scaffolds and CScrosslinked scaffolds
Prior to the fabrication of 3D PLCL/SF scaffolds (3DS), 2DS were cross-linked to improve the stability of membranes and submerged into 50 mL of mixed solution containing 50 mM 2-morpholinoethanesulfonic acid (MES) buffer, 30 mM EDC, and 8 mM NHS for up to 6 h. The cross-linked membranes were washed with deionized water three times and then dried in a vacuum oven. Subsequently, 2DS were immersed into foaming agent (concentration: 0.5 M NaBH 4 solution) and removed from the medium at predetermined time point. After rinsing three times with the deionized water and freeze-drying, the dried porous 3DS was obtained. The production of hydrogen bubbles from NaBH 4 solution is based on the Eq. (1) (Eq. (1)): To obtain the CS-crosslinked 3D scaffolds (3DCSS), the same cross-linking agent as mentioned above was rstly prepared with the only difference that 1 g of CS was added into the mixed crosslinking agent. 2DS were immersed into the cross-linking medium containing CS for 6 h. Subsequently, free CS and salt on the 2DS were washed with deionized water for three times. After drying, the CS cross-linked 2D membranes were treated with the above gas foaming method and subsequently processed (rinsed and freeze-dried) to afford 3DCSS.
For preparing 3D gas-foamed nano ber scaffolds with precisely controllable thickness, a mold with predetermined height (3 mm) was employed. Brie y, a concave mold made of hot-poured Te on was used as a base, and a matching-sized glass was then covered and adhered to the base by using light-curing glue. The electrospun nano ber mat was placed at the center of the mold and immersed into the NaBH 4 solution as described above. After half an hour, the holder was removed and rinsed thrice with deionized water. After freeze-drying, 3DS or 3DCSS with precise thickness were obtained.

Characterization
The bulk volume of samples was measured by ethanol displacement method [18]. The volume percentage of specimens was then calculated by dividing volume of the gas foamed scaffold at different time intervals with the volume of the initial membrane. The bulk density of specimens was calculated by dividing the specimen mass with the bulk volume. The porosity of the nano ber membranes and expanded scaffolds was measured by the liquid displacement method and calculated according to where P is the porosity, Ws is the weight of the sample after soaking for up to 10 min in ethanol with a density of 'r', Wd is the weight of the dry scaffold, and V is the volume of the sample (n = 3).
The photographs of 2D and 3D gas-foamed scaffolds were recorded by a digital camera. Samples were mounted on an aluminum stub with carbon tape and then sputter-coated with Au. Scanning electron microscopy (SEM, Phenom XL, Phenom Scienti c Instruments Co. Ltd., Shanghai, China) was then used to observe the morphology of different scaffold. Subsequently, the surface morphology of nano bers and cross-sections of different samples were imaged at an accelerating voltage of 10 kV. SEM images were analyzed by using Image J software to measure the pore area and gap distance of scaffolds. For pore area measurement (n = 20), the side length of the different shaped (triangular and diamond) holes was rst measured, and then the area was calculated.
The elements on the surface of 3DCSS were analyzed by an energy dispersive spectrometer (EDS, JSM-7500F, China). Structural elucidation of scaffolds was carried out by using Fourier transform infrared spectroscopy (FTIR) by using a Nicolet-6700 FTIR spectrometer (Thermo Fisher Scienti c, USA) in the range of 3800-600 cm − 1 .
The antioxidant activity of different samples was assessed by 2,2-diphenyl-1-picrylhydrazyl (DPPH) assay. Brie y, 10 mg of each sample was mixed with 3 mL of 0.1 mM DPPH/ethanol solution, while protected from the light at room temperature. After reacting for 0.5 h, the absorbance of each mixture was measured at 517 nm using ethanol as the blank. Pure ascorbic acid (AA) was used as a positive control. Whereas, DPPH solution without samples was taken as a blank. Radical scavenging activity was expressed as the scavenging rate and calculated by using the following Eq. (3): Where A is the DPPH· scavenging rate; A B and A S are the absorbance of the blank and experimental samples, respectively.
The water absorption capacity of different samples was determined according to the previous report [28]. The dry sample with known weights (w d ) was added into a 25 mL ask containing 20 mL of phosphate buffered saline (PBS, pH = 7.4) solution at room temperature for up to 2.5, 5, and 15 min. Afterwards, the excess water was absorbed from the scaffolds with a lter paper and the weight of the scaffold was recorded (w w ). The water absorption rate (w) was calculated according to Eq. (4): The mechanical properties of scaffolds, including uniaxial tensile testing and compressive testing of samples were measured at room temperature by using a universal materials tester (Instron-5542, Canton, USA). Rectangular-shaped specimens of 2DS (40 mm × 10 mm × 0.5 mm), 3DS (40 mm × 10 mm × 3 mm) and 3DCSS (40 mm × 10 mm × 3 mm) were evaluated by tensile testing in wet state. A cross-head speed of 5 mm/min was used for all of the specimens examined until breakage point. For the compression test, rectangular-shaped samples with a side length of 10 mm and a thickness consistent with that described above were evaluated at a cross-head speed of 1 mm/min. The stress-strain curves of specimens were drawn by using the data recorded by the machine. Ultimate tensile strength (UTS) and elongation at break (Eb) were determined. The Young's moduli (E) and compressive moduli of scaffolds were calculated analytically by using a slope tting method of the initial linear region of the stress-strain curves.

Evaluation of the biocompatibility of scaffolds in vitro
Articular cartilage was derived from New Zealand white rabbit with an approval obtained from the Shanghai Pulmonary Hospital Ethics Committee and chondrocytes were isolated as previously reported [29]. The acquired chondrocytes were cultured, and expanded in the culture medium (high glucose, Dulbecco's modi ed Eagle's medium (DMEM, Gibco, USA) supplemented with 10% fetal bovine serum (FBS, Hyclone, USA) and 1% penicillin/streptomycin/amphotericin B solution in an incubator at 37°C [30].
Chondrocytes at the second passage were used for the subsequent in vitro evaluations.
After sterilization by ultraviolet (UV) light for up to 6 h, 2DS, 3DS and 3DCSS were xed in the bottom of the 24-well plate, and chondrocytes at a density of 3.0 × 10 4 cells in 500 µL of medium were evenlyseeded onto each sample. After culturing for up to 24 h, scaffolds were washed with PBS thrice and treated with live/dead staining kit (In vitrogen, USA). After washing, the uorescence of cells was observed by using a confocal laser scanning microscope (Nikon, A1RMP, Japan).
The cell seeding e ciency of different scaffolds was analyzed as described previously [31]. The same cell density as above was evenly-dropped onto the materials. After an incubation for up to 4 h, the media from different samples containing unseeded cells were collected and counted (lost cell number). The cell seeding e ciency of different scaffolds was calculated based on Eq. (5): The proliferation of chondrocytes on scaffolds was determined by a cell counting kit 8 (CCK-8) assay.
Chondrocytes (3.0 × 10 4 ) were seeded on 2DS, 3DS, and 3DCSS for up to 1, 3, and 7 days in a 24-well plate. At each time point, the medium was removed and the scaffolds were washed with PBS thrice. After that, 200 µL of the CCK-8/DMEM mixture (1:10 v/v) was added to each well and incubated for up to 2 h at 37 ℃, the absorbance was measured at the wavelength of 450 nm.
To observe the cell morphology and cell in ltration at pre-determined time points, chondrocytes grown on 2DS, 3DS and 3DCSS were xed and dehydrated with an ascending series of graded ethanol (30-100 %).
The morphology of chondrocytes was examined by SEM. Dehydrated cell seeded samples were para nembedded, sectioned (5 µm) and stained by hematoxylin and eosin (H&E). Chondrocyte in ltration was imaged by an optical microscope (Leica Microsystems, Germany).
The feasibility of scaffolds in promoting chondrogenesis in vitro was further evaluated by combining them with auricular chondrocytes. Firstly, 200 µL of chondrocyte suspension at a concentration of 3.0×10 8 cells/mL was seeded onto 2DS (diameter, 6.5 mm and thickness, 0.5 mm), 3DS and 3DCSS (dimeter, 6.5 mm and thickness, 3 mm). Subsequently, these cell-scaffold constructs were incubated at 37°C under 5% CO 2 for 2 h and then cultured in pre-warmed chondrogenic medium for up to 6 weeks in vitro. The chondrogenic medium used in this article is consistent with that of the previous study [32], and changed every 3 days during the culture duration. After 6 weeks, the chondrocyte-scaffold constructs were xed in 4 % Paraformaldehyde (PFA) and stained with H&E and Safranin-O for histological analysis and assessment of the secreted matrix of the engineered neocartilage tissues. Collagen type II was detected by immunohistochemical staining by following a previous method [33]. The specimens (n = 3) were minced to perform cartilage-related biochemical evaluations for DNA, glycosaminoglycan (GAG), and total collagen content as quanti ed by the PicoGreen dsDNA assay (Invitrogen), dimethylmethylene blue assay (Sigma-Aldrich), and hydroxyproline assay (Sigma-Aldrich) [34]. To determine the compressive moduli of cell-scaffold constructs (n = 3), a biomechanical testing machine (Instron-5542, Canton, USA) was used. Samples were subjected to uncon ned compression tests at a strain rate of 1 mm/min until they reached 50 % compressive strain. The compressive moduli of tested samples were calculated based on the slopes of the generated stress − strain curves.
To explore the changes in the in ammatory response and compare the chondro-protective ability of different scaffolds, chondrocytes were seeded on 2DS, 3DS, and 3DCSS in a 6-well plate with a density of 3.0 × 10 5 cells/well and treated with interleukin-1 beta (IL-1β, 10 ng/mL) for up to 24 h. Real-time quantitative polymerase chain reaction (RT-qPCR) was performed to analyze the expression of proin ammatory genes, including tumor necrosis factor-alpha (TNF-α), IL-1β, matrix metalloproteinase-13 (MMP13). Cell-laden scaffolds (n = 3) were ground within TRIzol reagent (Invitrogen, Life Technologies) to extract total mRNA. About 1 µg of RNA was used for cDNA synthesis by M-MLV cDNA Synthesis Kit (Promega) according to the manufacturer's instruction. RT-qPCR was performed by using SYBR Green PCR Master Mix (Enzynomics) and ABI StepOnePlus™ Real-Time PCR system (Applied Biosystems). All primer sequences are listed in Table S1 (supplementary information). The expression of all genes was normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and to gene expression of chondrocytes before seeding into the different samples and the relative expression was calculated by − 2 ΔΔCt method.

Assessment of articular cartilage regeneration in rabbits
All animals were performed according to the standard guidelines approved by the ethics committee of Shanghai Jiao Tong University (SJTU). To determine the articular cartilage regeneration capability of engineered scaffolds, 2DS, 3DS, and 3DCSS were prepared prior to surgery. Adult healthy male New Zealand white rabbits, 4 months old and weighing approximately 2.5 kg were randomly selected and divided into three groups (n = 4 per group). After anesthesia with 10 % of chloral hydrate, cartilage defects (4 mm in diameter and 3 mm in depth) were created by using a stainless-steel punch on the trochlear groove of the distal femur. 2DS (diameter, 4 mm and thickness, 0.5 mm) and matching 3DS as well as 3DCSS (diameter, 4 mm and thickness, 3 mm)) were implanted into the defects. Finally, surgical incisions were closed with sutures. All the rabbits were sacri ced at 12 weeks to harvest the knee joint.
Then, the harvested rabbit articular cartilages were rstly observed and scored according to International Cartilage Repair Society (ICRS) macroscopic scoring standard. Then, cartilage samples were decalci ed, sectioned, and characterized for histological and further immunohistochemical analysis including H&E, Safranin-O/ fast green (Safranin-O/FG), and collagen type I/II staining [33]. Furthermore, the regenerated cartilage was also assessed with Modi ed O'Driscoll histological scoring and Mankin score. The grading and scoring criteria are shown in the Table S2 − S4 (supplementary information). Immuno uorescence staining for IL-1β and TNF-α was performed on sectioned slices of different groups and the procedures was the same as in Sect. 2.5. The synovial uid was collected by using a 2 mL syringe with an 18-gauge needle and centrifuged at 4000 rpm for 20 min at 4°C. The supernatants were collected and frozen at -80°C. IL-1β and TNF-α were assayed by enzyme-linked immunosorbent assay (ELISA), Rabbit IL-1 ELISA Kit and Rabbit TNF-α ELISA Kit; Abcam) according to the manufacturer's instructions.

Statistical analysis:
Experimental data are presented as mean ± standard deviation (SD). For statistical signi cance, Student's t test or one-way analysis of variance (ANOVA) with Tukey's post-hoc test was performed where appropriate con dence level. A p values < 0.05 was considered to be statistically signi cant (*p < 0.05, **p < 0.01, ***p < 0.001).

Fabrication and characterization of 2DS, 3DS and 3DCSS
According to the schematic diagram shown in Fig. 1, the fabrication process of 2DS, 3DS, and 3DCSS scaffold can be mainly divided into the four steps: (1) conjugate electrospinning, (2) cross-linking, (3) gas foaming in a con ned mold, and (4) freeze-drying. The fabrication of 3D scaffolds is simple and e cient, which is envisaged to be applied to the large-scale industrial process.
After treatment with NaBH 4 solution for 5 min, the original mats were gradually expanded and the thickness of 3DCSS increased faster than that of the 3DS ( Fig. 2A − B). The thickness of 3DS increased from 0.5-4 mm (Fig. 2C), while it increased from 0.5-8 mm for 3DCSS (Fig. 2D). The effect of the gas foaming time on the volume expansion, porosity, and density of the expanded scaffolds was investigated in detail. As shown in Fig. 2E, after 5 min of gas foaming, the volume of the 3DS and 3DCSS was 8 and 18 times larger than that of the unexpanded membranes. The bulk density of the 2D mats was 0.27 g/cm 3 , which decreased to 0.03 and 0.02 g/cm 3 for 3DS and 3DCSS, respectively (Fig. 2F). Similarly, the porosity increased from 71-92 % and 69-97 % for 3DS and 3DCSS after gas foaming, respectively (Fig. 2G). To precisely control the thickness of the 3D scaffold, a mold consisting of a Te on concave substrate assembled with a matching-sized glass piece was used. Due to the limitation in the height of the mold, the 3D scaffolds stopped expanding when they reached a certain thickness during the foaming process. After freeze-drying, a 3D nano ber scaffold with a stable and pre-de ned thickness was obtained (Fig. 3A). The direction of alignment of scaffolds is shown in Fig. 3B, in which the red arrow indicates the alignment direction of the nano bers. Figure 3C exhibited the gross appearance of 2DS with a thickness of 0.5 mm. After expansion in the mold, 3DS and 3DCSS with precise thickness (3 mm) were obtained and the photographs were shown in Fig. 3G and Fig. 3K, respectively. 2DS showed compactpacked structure of the cross-section (Fig. 3D − E) and densely aligned nano bers and yarns of the surface (Fig. 3F). After expansion, good interconnectivity and continuous layered structure was achieved in 3DS (Fig. 3H − I) and 3DCSS (Fig. 3L − M). In addition, a loose surface with preserved aligned nano brous topographical cues was formed in 3D gas-foamed scaffolds (Fig. 3J & N).
The elements on the surfaces of 3DCSS were analyzed by SEM equipped with EDS. The sulfur elements contained in the CS were evenly distributed on the ber surface of the 3DCSS (Fig. 4A − C). The composition of different samples was further investigated by FTIR and the spectra in the range of 3800-  [36,37]. Moreover, an additional carbonyl peak at 1635 cm − 1 in the 3DCSS spectrum near amide I further indicated the presence of CS in the 3DCSS. Taken together, the FTIR spectra con rmed the presence of the incorporated components into the scaffolds as well as their crosslinking with the CS.
The antioxidant activity of PLCL, 2DS, 3DS and 3DCSS was analyzed by using DPPH radical scavenging assay. Ascorbic acid was taken as a standard antioxidant agent. The results were shown in Fig. 4F. The antioxidant activity was found to be 82. 49 (Fig. 4L), and the compressive modulus was found to be 1.23 ± 0.26 kPa and 1.99 ± 0.16 kPa for 3DS and 3DCSS, respectively (Fig. 4K). The maximum water absorption of 3DS and 3DCSS scaffolds (≈ 1800 %) was reached within 5 minutes, and was signi cantly higher than that of the 2DS (≈ 500 %) (Fig. S2).

In vitro cytocompatibility studies
Live/Dead staining assay was used to detect cell viability in the constructs. Both 2D and 3D gas-foamed scaffolds showed minimal cell cytotoxicity with well-proportioned distribution of chondrocytes after 24 h of culture under a confocal microscope. More chondrocytes survived on the 3D gas-foamed scaffolds than that of the 2DS groups. The morphology of attached chondrocytes at day 5 was observed by SEM. Chondrocytes integrated well with the scaffolds in all three groups and proliferated along the nano bers ( Fig. 5D − F). Cell in ltration, assessed by H&E staining, indicted more numbers of cells in ltrated into the 3D gas-foamed scaffolds compared to 2DS (Fig. 5G − I). Moreover, 3D gas-foamed scaffolds presented considerably better cell seeding e ciency after 4 h of culture than 2DS (Fig. 5S).
Cell proliferation was determined by CCK-8 assay, which demonstrated that all nano ber scaffolds had good cell viability in vitro (Fig. 5T). On days 1 to 7, the proliferation of chondrocytes on the 3D gasfoamed scaffolds was signi cantly higher than of the 2DS. Notably, at day 7, the 3DCSS showed highest numbers of cells among all of the investigated groups. These results reveal that 3DCSS could provide a conducive environment for cell growth and in ltration in vitro, which may also have implications for the in vivo applications.

In vitro evaluation of cartilage regeneration
To further validate the regenerative ability of engineered scaffolds in vitro, chondrocytes were evenly dropped onto different scaffolds and grown for up to 6 weeks. All of the scaffolds, maintained their circular shapes and presented cartilage-like tissue development surrounding the construct (Fig. 5J − L).
Histological analysis demonstrated that all of the groups contained the cartilage-speci c ECM deposition (Fig. 5M − O). 2DS and 3DS specimens exhibited positive staining for Safranin-O and collagen type II only in some regions of the regenerated cartilage ( Fig. 5P-Q & Fig. S3A-B). Intriguingly, 3DCSS exhibited a relatively dense and homogeneous distribution of cartilaginous tissues than that of the other groups ( Fig. 5R & Fig. S3C). Moreover, the quantitative analysis showed that the compressive modulus, DNA content, total GAG content, and collagen content were higher in 3DCSS as compared to the other groups, indicating the 3DCSS were better for the regeneration of matured cartilage as well as cartilage-speci c matrix production (Fig. 5U-X).

In vitro evaluation of anti-in ammatory effect
To con rm the chondro-protective ability of different scaffolds, chondrocytes were seeded on scaffolds and treated with IL-1β to mimic the in ammatory microenvironment after injury. Immunostaining results showed the expression of IL-1β and TNF-α was more prominent in 2DS and 3DS groups compared to the weak staining of the 3DCSS group (Fig. 5Y). The in ammatory gene expressions were also evaluated by RT-qPCR (Fig. 5Z). As anticipated, the in ammatory genes, such as IL-1β, TNF-α and MMP13 were only marginally expressed in the 3DCSS group than that of the other groups, con rming the better antiin ammatory effect of 3DCSS in vitro.

In vivo evaluation of articular cartilage regeneration
Cartilage regeneration in vivo was a crucial criterion to determine the appropriateness of the scaffold for CTE. A rabbit articular cartilage defect model was used to assess the therapeutic e cacy of 2DS, 3DS and 3DCSS. After implantation for up to 12 weeks, the gross appearance of retrieved articular tissues was observed and compared. The cartilage defects in the 2DS groups were clearly visible, indicating the poor integration of the scaffolds with the normal cartilage tissues (Fig. 6A). In contrast, the defects in the 3DS group were only partially covered by the newly-formed cartilage-like tissues (Fig. 6B). Notably, 3DCSS group achieved a best regeneration outcome, as the defect was lled with neo-cartilage, and the boundary between defect site and surrounding cartilage was almost invisible (Fig. 6C).
H&E staining and Safranin O-fast green staining assays of the retrieved cartilage samples were further performed to examine the formation of brous tissue, and the deposition of ECM. In 2DS groups, an irregular border between the neo-tissues and native cartilage was observed and only brous tissues were detected in the joint, where only a few unevenly distributed cartilage clusters lled the defect (Fig. 6D &  D1). For the 3DS, a thin layer of cartilage-like tissue was observed on the surface of the defect region, whereas some brous tissues containing fewer cells were found (Fig. 6E & E1). By contrast, the 3DCSS groups displayed a well-integrated boundary lled with homogeneously distributed newly formed tissues featured with the chondrocyte-speci c lacunas (Fig. 6F & F1).
2DS group also lacked Safranin O-fast green staining in the defect region (Fig. 6G & G1). For 3DS groups newly formed tissue with a weak staining for the Safranin O-fast green lled the defect region (Fig. 6H &   H1). These results indicated that 2DS and 3DS groups exhibited brous tissue, which lacked the deposition of the cartilage-speci c ECM. By contrast, in 3DCSS groups, it was shown that the defect area was covered with a large amount of positively-stained cartilage ECM, indicating the formation of matured cartilage as well as the deposition of the abundant GAGs (Fig. 6I & I1). Consistent with the histological staining trends, the immunohistochemical analysis of the collagen type II exhibited a signi cant deposition of the collagen type II (Fig. 6L & L1) in 3DCSS groups compared to only a limited and slight deposition of collagen in the 2DS (Fig. 6J & J1) and 3DS groups (Fig. 6K & K1). Moreover, the 3DCSS groups showed only a weak staining for the collagen type I than that of the 2DS and 3DS groups. The collagen type I seemed to be the dominant form in the newly formed tissues in the defects treated with the 2DS as well as a few regions of 3DS groups (Fig. S4). All of these results demonstrated that 3DCSS achieved a satisfactory regeneration of cartilage along with the accumulation of the cartilage-speci c ECM rather than the brocartilage formation.
The intra-articular in ammatory response was discerned by immuno uorescent staining for different types of in ammatory markers, which demonstrated a high expression of IL-1β and TNF-α in the 2DS group as compared to the other groups. Similar to the native cartilage, weak uorescent expression of pro-in ammatory markers was observed in 3DCSS groups (Fig. 7A). Meanwhile, quanti cation results of in ammatory factors, such as IL-1β and TNF-α, showed the less expression of in ammatory markers in the 3DCSS groups compared to the 2DS and 3DS groups, indicating that 3DCSS attenuated the in ammatory response and provided conducive environment for neo-cartilage formation ( Fig. 7B-C).

Discussion
Electrospinning, as a universal and cost-effective nano ber fabrication technology, has been extensively explored for developing tissue engineering scaffolds and devices for a myriad of clinical and industrial applications [38]. However, it is known that 2D nano ber membranes generated by conventional electrospinning pose certain restrictions for cell in ltration due to their tightly-packed layers and dense super cial pores, which lack 3D-like architecture for tissue regeneration. Recent studies reported numerous attempts to afford 3D tissue-engineered scaffolds with nano brous structures based on 3D printing or self-assembly of short nano bers as well as self-folding nano ber mats [31,39,40]. Besides, it has been reported that the porosity and pore size of electrospun scaffolds can be tailored by varying the ber size [41,42]. These technologies are limited by some constraints, such as exposure of the scaffolds to the toxic solvents, lengthy and time-consuming preparation procedures as well as the requirement of the special instrumentation. Alternatively, the gas foaming technique, has been frequently exploited for the post-treatment of electrospun nano ber membranes to afford 3D nano ber scaffolds. Gas foamed scaffolds often feature laminated brous layered structure with high porosity and interconnected pores, providing su cient space and a stable 3D-like architecture for cell in ltration and proliferation [15,17]. However, to the best of the authors knowledge, most of the gas-foamed scaffolds have been assembled from synthetic polymers, such as poly(caprolactone) (PCL), poly(L-lactide-co-ε-caprolactone) (PLCL), or poly(L-lactide) (PLLA), which lack cell recognition cues, and necessitate post-modi cation with ECMderived proteins or peptides for tissue repair applications. Moreover, merely by adjusting the physical parameters, such as time or the concentration of the foaming agent to afford scaffolds with the controlled structure and size may not hold considerable potential to afford uniform expansion throughout the scaffold, especially in the central region [19,43]. To overcome these limitations, herein, we have realized bio-hybrid scaffolds composed of natural and synthetic polymers, namely, SF and PLCL which exhibited good hydrophilicity and biocompatibility. By cross-linking with CS and gas-foaming, the cell in ltration and pro-chondrogenic potential of expanded porous scaffolds was augmented leading to the better cartilage regeneration both in vitro and in vivo.
Appropriate choice of the materials is the basis for the successful preparation of 3D nano ber scaffolds. PLCL, a synthetic copolymer with mechano-elastic and biodegradable properties, imparts stable mechanical structure and extensibility to the hybrid scaffold; however, it lacks effective cell recognition sites due to its hydrophobic nature [44]. Silk broin, a protein-based natural polymer may contain ligands that can be recognized by the cell-surface receptors, which may improve the biocompatibility and reduce the regulatory constraints [7,45]. As a favorable natural materials, SF exhibits unique properties, such as good biocompatibility and biodegradability, which may also increase the hydrophilicity of the scaffold materials [46]. Apart from the aforementioned advantages, SF is less risky in terms of the infection compared to the other proteinaceous materials [11]. Hence, a blend of PLCL and SF was employed for the fabrication of scaffold, which has already been exploited for the regeneration of other types of tissues, such as tendons, skin, blood vessels, and nerves [47]. To further improve the bioactivity of the scaffolds, various types of strategies have been proposed to modify the scaffolds. Kartogenin, a small molecule drug that promotes cartilage lineage commitment of stem cells, has been conjugated with amphiphilic polyurethanes nanoparticles for leveraging the chondrogenic effect, but its oil solubility predisposes it to the exposure to the toxic organic solvents during the modi cation process [48]. Cartilage-decellularized matrix, has recently been considered as a chondrogenic material to be incorporated with synthetic degradable polymers for inducing chondrogenesis [42]. However, it is di cult to be homogenized or to be dissolved in organic solvents, making it an impossible material to be processed into biomimetic ECM-like structures on the nanoscale [29]. Besides, there are still immunological risks due to their exogenous sources from porcine or goat.
In this work, CS, the composition found in cartilage ECM was evaluated to improve the synthesis of the cartilage-speci c matrix and to modulate the in ammatory microenvironment after injury [49].
Chondroitin sulphate, a negatively-charged polysaccharide broadly distributed in the cartilage, is involved in various physiological activities, such as cell growth and differentiation as well as the regulation of the joint function [50,51]. However, the weak mechanical properties of CS may limit its applications for CTE.
More targeted design is suggested to incorporate it with the other active substances for designing engineering cartilage scaffold [52,53]. In this study, CS cross-linked PLCL/SF scaffolds were successfully prepared by using EDC/NHS coupling chemistry, which exhibited considerable potential for cartilage regeneration in vitro and in vivo (Fig. 1). However, the introduction of CS increased the hydrophilicity of the scaffolds, leading to a more rapid increase in the thickness during gas foaming process, making it a challenge to develop 3D electrospun nano brous scaffolds with controllable size (Fig. 2). To achieve a precise control of 3D electrospun scaffolds, various approaches have been put forwarded. By using a conductive mold as a collector, Sun et al. [54] fabricated a 3D nanoyarn scaffold with hexagonal shape, which were mineralized for bone regeneration. Chen et al. [55] employed short nano ber powder as bioinks for 3D printing of scaffolds with controllable pore size. Other advanced approaches, such as origami-and kirigami-based techniques may offer a modality by which electrospun membranes may be transformed into 3D structures [41]. Although the aforementioned studies proved to be a signi cant advancement in fabricating 3D scaffolds with accurate size, the super cial pores with lacking porous layered structures may compromise these approaches. In this study, by considering the cost and reproducibility, a simple mold made of a combination of a Te on and glass was employed, which is not only easy to be assembled, but is also chemically inert, avoiding interference with the foaming process. After gas foaming by using this "con ned expansion" molding technology, 3D scaffolds with customized thickness and multi-layered structure were generated (Fig. 3).
The high porosity is a fundamental characteristic that enables scaffolds to provide adequate space for cell in ltration. During the gas foaming process, bubbles converge within the nano ber pores and exert pressure on the surrounding bers, resulting in an increase in the porosity, which is macroscopically manifested as an increase in the volume and thickness of scaffolds. A highly porous scaffold with an interconnected pore network is required for uniform spatial distribution of cells and to minimize the diffusion-limiting effects on nutrients and waste products [45]. This is particularly essential for highdensity chondrocyte culture in vitro to produce clinically signi cant volume of cartilage tissues. Furthermore, high porosity and aligned topographical cues facilitate e cient intercellular contact, which has been associated with the high rates of the ECM synthesis [56]. After treatment with the foaming agent and freeze-drying, the 3D gas-foamed scaffolds retain their aligned nano ber morphology with higher porosity, indicating their better performance over the nano brous membranes in promoting cell in ltration and cartilage-related ECM secretion. Furthermore, an ideal scaffold should be able to withstand the speci c strength. The mechanical properties of 3DCSS were signi cantly better than that of 3DS in both tensile and compressive tests (Fig. 4), which may be due to the stronger hydrogen bonding between the CS as well as the formation of amide bonds between the amino groups and the carboxylic groups.
The biocompatibility and bioactivity of scaffolds are important issues which strongly affect the feasibility of cartilage regeneration. The current results showed that chondrocytes stably survived on all of the scaffolds, indicating good biocompatibility and low cytotoxicity of both 2D membranes and 3D gas-foamed scaffold. Notably, 3DCSS achieved most rapid proliferation of chondrocytes as compared to the other groups and formed uniform cartilage-like tissues along with the secretion of the cartilaginous matrix both in vitro and in vivo. The biochemical results of chondrocytes cultured on different samples reemphasized the favorable in uence of 3DCSS groups on the chondrogenic phenotype in vitro (Fig. 5). This is ascribed to the reason that the 3DCSS scaffolds mimic the GAG-rich ECM of chondrocytes, which may provide the cultured cells with a niche-like microenvironment, allowing them to maintain the differentiated chondrogenic phenotype in vitro [57]. Although the mechanical properties of the scaffolds inevitably declined after gas foaming owing to the increased porosity and a loose structure, the 3DCSS showed improved mechanical parameters than that of the 3DS. This may help to present a stable 3D environment for the growth of chondrocytes and cartilage tissue regeneration. In addition to the improvements of the mechanical properties, the introduction of CS also played a crucial role in enhancing the bioactivity of scaffold. The precise role of CS in creating a chondro-inductive environment for the neocartilage regeneration is not very clear. Previous studies have shown that chondrocytes cultured on CSmodi ed chitosan membranes retained their phenotype and produced cartilage-speci c matrix [57]. In another study, the introduction of CS downregulated the expression of genes encoding proteolytic enzymes involved in cartilage degradation, suggesting that CS may exert both chondro-protective and anti-in ammatory effects, which may prove to be bene cial for CTE applications [58]. Consistent with these previous results, 3DCSS scaffolds showed the bioactivity in promoting the maturation of chondrocytes and secretion of the cartilage-speci c ECM as compared to the CS-free groups (Fig. 6). The existence of the CS might have multiple functions, such as the sequestration of the growth factors or direct interaction with the cells. The interaction of growth factors (GFs) with anionic domains in GAG, of which CS is a primary component, is well-known to play an essential role in morphogenesis and tissue homeostasis [59]. Growth factor-ECM interactions has been previously reported to prolong the half-life and activity of growth factors [25]; therefore it can be assumed that the sulfate domains of CS may interact with some of the GFs, thereby activating the required signaling that favor cartilage regeneration. Furthermore, it has been reported that CS possibly stimulates the synthesis of proteoglycans by inhibiting the synthesis of proteolytic enzymes and other factors that cause the cell apoptosis and cartilage matrix degradation [60]. This would explain the better potential of the 3DCSS scaffolds over other CS-free groups in reducing the synthesis of pro-in ammatory cytokines, such as IL-1β and TNF-α (Fig. 7), thus attenuating in ammation and promoting cartilaginous tissue regeneration after the occurrence of joint injury.

Conclusion
In summary, we described a novel 3D CS-crosslinked biomimetic porous nano brous scaffolds with precise thickness via gas foaming in a pre-designed mold and subsequent freeze-drying for promoting cartilage regeneration and moderating joint in ammation. 3D gas-foamed scaffolds demonstrated low density, appropriate porosity and fast water absorption. Moreover, they preserved nano-topographical architecture with multilayered structures and stable mechanical properties. All of the scaffolds exhibited good cytocompatibility, while 3D gas-foamed scaffolds promoted seeding e ciency and proliferation of chondrocytes than that of 2DS. More importantly, biomimetic 3DCSS displayed good biological outcomes than those of the other groups as evidenced by the formation of cartilage-speci c ECM, signi cant regeneration of articular cartilage in a rabbit model and less expression of pro-in ammatory factors both in vitro and in vivo. Moreover, the better mechanical and biological properties of 3DCSS were effective for the regeneration of the articular cartilage. Taken together, this strategy of designing biomimetic scaffolds as well as improving their performance by gas foaming may have broad implications for CTE applications.
Declarations Figure 1 Schematic illustration of the preparation of 2DS, 3DS, and 3DCSS scaffolds for CTE. 2DS were fabricated by conjugate electrospinning, crosslinked via CS, and expanded by using gas foaming. The bioactivity of membranes was assessed in vitro and in vivo. *p < 0.05, **p < 0.01, ***p < 0.001.  . RFI = Relative fold induction. Each value represents the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001. staining images of the repaired cartilage defects in different groups at 12 weeks after surgery. The blue arrows point toward the intact cartilage area, and the red arrows indicate the cartilage defect area. ICRS macroscopic assessment score (M), modi ed O′Driscoll histological score (N) and Mankin score (O) of the repaired tissue. Each value represents the mean ± SD (n = 4). *p < 0.05, **p < 0.01, ***p < 0.001. Figure 7 (A) Immuno uorescent staining for IL-1β and TNF-α of the defect regions 12 weeks post-surgery.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementaryinformation.docx