Microfabrication of a biomimetic arcade-like electrospun scaffold for cartilage tissue engineering applications

In recent years, the engineering of biomimetic cellular microenvironments has emerged as a top priority for regenerative medicine, being the in vitro recreation of the arcade-like cartilaginous tissue one of the most critical challenges due to the notorious absence of cost- and time-efficient microfabrication techniques capable of building 3D fibrous scaffolds with precise anisotropic properties. Taking this into account, we suggest a feasible and accurate methodology that uses a sequential adaptation of an electrospinning-electrospraying set up to construct a hierarchical system comprising both polycaprolactone (PCL) fibres and polyethylene glycol sacrificial microparticles. After porogen leaching, the bi-layered PCL scaffold was capable of presenting not only a depth-dependent fibre orientation similar to natural cartilage, but also mechanical features and porosity proficient to encourage an enhanced cell response. In fact, cell viability studies confirmed the biocompatibility of the scaffold and its ability to guarantee suitable cell adhesion, proliferation and migration throughout the 3D anisotropic fibrous network during 21 days of culture. Additionally, likewise the hierarchical relationship between chondrocytes and their extracellular matrix, the reported PCL scaffold was able to induce depth-dependent cell-material interactions responsible for promoting a spatial modulation of the morphology, alignment and density of the cells in vitro.


Introduction
One key design criteria of the next generation tissue engineering (TE) scaffolds is the accurate recreation in vitro of the extracellular matrices (ECMs) 3D hierarchical complexities, including the mimicking of the architectures, biochemical gradients and mechanical properties of the native tissues [1,2]. This purpose is particularly evident in cartilage TE applications due to the challenges associated with the engineering of an arcade-like arrangement similar to the cartilaginous collagen network, where the orientation of the fibrils progress from perpendicular relatively to the subchondral bone surface in the deepest region-the pillars of the arcades-, to random in the middle zone, to parallel in the superficial zone-the arches of the structure [3,4]. The successful replication of these depth-dependent topographic cues is mandatory to modulate an enhanced cell response since the chondrocytes are capable of adapting their morphology and biochemical expression patterns according to the geometry of the surrounding microenvironment [5]. Indeed, considering that the natural regeneration process of articular cartilage is impaired by the combination of intrinsic factors such as low cell density, poor chondrocyte proliferation and absence of vascularisation, a clinically fitted biomimetic scaffold could be a crucial tool to replace the damaged cartilaginous tissue and consequently restore the functionality of the injured area [6].
Following this trend, during the past few years, a very impressive set of advanced scaffolds targeting cartilage repair has been reported, presenting not only promising results regarding chondrogenesis, but also pointing out complementary TE parameters like bioactivation pathways, cell culture conditions and, principally, groundbreaking design and fabrication methodologies to build 3D cellular microenvironments [7][8][9]. As a matter of fact, recent studies have pointed out that hydrogels loaded with cells can be presented in form of microgels or bioinks with the purpose of being assembled or bioprinted, respectively, into 3D constructs capable of boosting cell response [10,11]. Other cartilage scaffolding approaches allow the possibility of engineering porous systems with pore size and/or orientation able to mimic the depth-dependent biochemical and biomechanical properties of natural cartilage, leading to a spatially controlled chondrocyte behaviour both in vitro and in vivo [12,13]. Although hydrogels and porous scaffolds can present interesting hierarchical architectures able to stimulate cartilaginous tissue regeneration, both designs are insufficient to guarantee topographical cues similar to their natural counterparts, where the fibrous morphology and orientation of the cartilaginous ECM are crucial to ensure structure and function. Thus, fibrous scaffolds have emerged as a relevant option for cartilage TE, bringing electrospinning to a preponderant position relatively to other microfabrication techniques (e.g. self-assembly and phase separation) due to its cost-effectiveness, reproducibility and versatility regarding important designing fundamentals such as fibre diameter (from nano to micro range), orientation and chemical composition [14][15][16]. However, since the conventional electrospinning setup is usually limited to the producing of 2D dense fibrous networks, there has been a growing focusing on adjusting this technology to match the requirements of a 3D biomimetic fibrous scaffold for cartilage TE applications. For example, by combining optimal technological parameters (e.g. solution viscosity, voltage, working distance and ambient conditions) with a predefined XY translation of the collector, Chen et al. were able to use a writing electrospinning modality to accurately fabricate an ultrathin cartilage mimetic scaffold capable of enhancing chondrocyte differentiation and directing the new tissue according to each simulated cartilaginous fibrous zone [17]. Other approaches suggested either the potential of processing electrospun fibres into inks for bioprinting highly porous systems suitable for inducing cartilage regeneration in vivo, or that a 3D fibrous/porous network could be fabricated via melt electrospinning and then incorporated into a soft hydrogel as reinforcing agent with the purpose of ensuring suitable anisotropic biomechanical properties for neocartilage formation [18,19]. Alternatively, using multi-layer electrospinning do not require complex equipment, leading to a simpler but time-consuming process to build 3D fibrous systems with controllable fibre size and orientation by simply adjusting the mandrel rotational speed. In fact, a recent work reported a five-layered electrospun scaffold with not only a hierarchical fibre orientation, but also with depth-dependent biological properties due to the sequential electrospinning of two types of collagen [20]. However, this type of scaffolds usually present low thickness and small pore size, limiting the success of cell infiltration and growth. In this regard, other suggested hypotheses to fabricate implantable functional cartilage TE scaffolds include: direct electrospinning into a 3D mould with the desired architecture, engineer composite scaffolds able to intercalate layers of electrospun fibres and porous decellularized extracellular matrix sponges, and perform sequences of freezing, freeze-drying and thermal treatment of an initial dispersion of electrospun fibres [21][22][23].
Taking this into account, since there is currently no feasible and efficient methodology able to guarantee the precise recreation of the arcade-like fibrous organisation of native cartilage, we suggest a simple four-step approach to develop a 3D biomimetic bi-layered electrospun scaffold with anisotropic features compatible with cartilage TE protocols. Indeed, based on the obtained results, we expect that the accessibility and scalability of the presented microfabrication technique could support and potentiate a wide range of cell culture protocols concerning cartilage regeneration.

Microfabrication of the bi-layered PCL electrospun scaffold
The PCL and PEG solutions were prepared by dissolving the polymers in a mixture of DCM:DMF (80:20, v-v) with a final concentration of 12% w/v and in chloroform with a concentration of 1.5 g mL −1 , respectively. The PCL solution was stirred at room temperature overnight while the PEG solution was prepared 45 min before the electrospinning process, including a 30 min period of stirring at a temperature of 50°C followed by a cooling period at room temperature. In this way, it was possible to guarantee the quality and stability of both polymer solutions, particularly the PEG solution since it was crucial to avoid its sol-gel transition before and during the electrospraying process. Figure 1 summarises the microfabrication procedure. Firstly (STEP 1 of Fig. 1), the PCL and PEG solutions were simultaneously electrospun and electrosprayed using an applied voltage of 25 kV, distinct flow rates (1 mL h −1 for PCL and 4.5 mL h −1 for PEG) and different needle tips (21 G for PCL and 18 G for PEG). Both the PCL microfibres and the PEG microparticles were collected in a rotating drum (width = 200 mm; diameter = 200 mm; 750 RPM) using a working distance of 15 cm. The final mesh presented a thickness of approximately 300 μm. A small rectangle (length = 4 mm and width = 3 mm) was cut from the PCL-PEG mesh and then rolled up inside a teflon mould to form the spiral shaped cylinder with defined dimensions (diameter = 3 mm and high = 3 mm) that is illustrated in the STEP 2 of Fig. 1. This spiralled cylinder was used as Bottom Layer (BL) for the final scaffold since it presented vertically aligned electrospun fibres relatively to the Top Layer (TL), which was fabricated according to STEP 3 of the Fig. 1. In this phase, the rotating drum was replaced by a static collector able to support the BL, making it a target for the newly formed PCL-PEG network with a thickness of approximately 500 μm. It is noteworthy to mention that during STEP 1 and STEP 3, the PEG solution was sequentially replaced by one fresh after each hour of processing in order to ensure its spinnability. The bi-layered PCL-PEG scaffold was finally subjected to a series of washes with distilled water at 37°C for 2 weeks in order to remove the sacrificial PEG microparticles and consequently increase the inter-fibre distance (STEP 4 of Fig. 1). The final anisotropic elctrospun 3D structure was freeze-dried and denominated PCL scaffold, presenting final dimensions (thickness = approximately 3.5 mm; diameter = 3 mm) that are compatible with the native articular cartilaginous tissue (2 mm to 4 mm of thickness) [24].

Characterisation of the PCL scaffold
A scanning electron microscope (SEM) Hitachi SU 70 (Hitachi High-Technologies) was used to evaluate the topographic features of the PCL scaffold. More precisely, the diameter and fibre-fibre distance together with the dimensions of the sacrificial microparticles were studied by directly analysing ten SEM pictures of the PCL-PEG network before and after the washing procedure described in the STEP 4 of Fig. 1. Similarly, a SEM analysis was conducted to investigate the anisotropic morphology of the PCL scaffold. Before the mechanical analysis, the swelling properties of the PCL scaffold were evaluated by immersing the samples (n = 5) into distilled water for 72 h at room temperature. The variation of the swelling ratio was calculated for different time intervals using the following formula: where R is the swelling ratio (mg mg −1 ), the W s is the weight of the swollen scaffold and W d the weight of the dried scaffold. The swollen PCL scaffolds were then subjected to compressive tests using a Shimadzu MMT-101 N (Shimadzu Scientific Instruments) with a load cell of 100 N. The compressive moduli of the samples were calculated by analysing the slope of the stress-strain curves in the linear region (R 2 ≥ 0.97 for all samples) after a pre-charge of 0.07 N. The samples were compressed until 15% strain with a compressive rate of 0.5 mm min −1 .

Cell viability studies
CP5 cells were maintained in DMEM/F-12 supplemented with 10% (v/v) FBS and 1% (v/v) Pen-Strep at 37°C in a humidified atmosphere of 5% CO 2 . The medium was refreshed three times a week and the cells were harvested at pre-confluence using trypsin/EDTA solution (0.25%). Prior to cell culture, the PCL scaffolds were sterilised in 70% ethanol for 3 h and then carefully washed with PBS before incubation in DMEM/F-12 medium for 1 h. Each scaffold was placed in a well of a 96-well plate and seeded with 0.5 × 10 6 CP5 cells suspended in 50 μL of culture media. The cell adhesion was guaranteed by incubating the PCL scaffolds for 2 h at 37°C and 5% CO 2 . Afterwards, fresh medium was added until a final volume of 200 μL per well was reached. The cell viability was evaluated via a non-toxic resazurin metabolic assay [25,26]. Briefly, at specific time points (day 1, 4, 7, 14 and 21), the scaffolds were incubated with fresh medium containing 10% of a resazurin solution (0.1 mg mL −1 in PBS) during a period of 4 h. Then, the resazurin reduction to resorufin was determined by spectrophotometry (Synergy™ HTX), more precisely, the absorbances at 570 and 600 nm were measured since these are the resazurin maximum absorbance and the resorufin maximum absorbance, respectively. The The elongation of the CP5 chondrocytes was studied by analysing both fluorescence and SEM pictures, where it was possible to accurate discern individual cells (n > 50 per sample). In detail, ImageJ software was used to determine the longer and smaller axes of the cells with the final goal of calculating the elongation factor (E) by the following formula:

Statistical analysis
Statistically significant differences were determined by using a one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test (Origin Software, *p < 0.05). Data are expressed as mean ± standard deviation.

Morphological and mechanical characterisation of the PCL scaffold
PCL was selected as bulk biomaterial for the microfabrication of the bi-layered anisotropic scaffold due to its well-known set of properties, particularly the remarkable biocompatibility/biodegradability and the versatile manipulation/stability features that allow an easy shaping towards dissimilar TE scaffolds (e.g. 2D films, electrospun fibres, 3D porous networks, etc.) [28][29][30]. As it is possible to see from the STEP 4 of the Fig. 1, the PCL scaffold presented a 3D arcade-like fibrous structure likewise its natural counterpart due to the perpendicularity established between the vertical alignment of the BL relatively to the horizontal orientation of the fibres located onto the TL. More precisely, this biomimetic fibre arrangement was successfully accomplished via a sequential microfabrication process that started by simply curling a rectangle cut from the initial PCL-PEG network to form a spiralled cylinder with the desired vertical fibre orientation. Afterwards, as the initial rotating drum was replaced by a static collector, it was possible to target the BL with the intention of fabricating a new fibremicroparticle system above it and, consequently, ensuring an orientation of 90°with respect to the deepest zone. For both TL and BL, PCL was electrospun simultaneously with the electrospraying of PEG with the purpose of, firstly, building a network of both fibres and sacrificial microparticles and, secondly, enabling the enlargement of the fibre-fibre distance after PEG removal. This tactic to increase the pore size of the PCL electrospun mesh is a simple variation of other reported studies in which the authors used the leaching of water-soluble materials located into the fibrous networks to enhance the inter-fibre distance, leading to an adaption of their porosity features [31,32]. Indeed, it was possible to optimise the inter-fibre distance without damaging the integrity of the 3D PCL scaffold nor disrupting the morphology of both layers, where the fibres presented a diameter of 1.5 ± 0.5 μm and maintained their depth-dependent orientation together with a smooth and defect free surface. Considering the design of the scaffold, it was important to target a fibre size in a micro range since the possible usage of nanofibres could have been a restrictive factor for cell penetration across the 3D anisotropic fibrous system due to the associated smaller pore size comparatively to microfibrous electrospun scaffolds [33]. Figure 2a compares the pore size (fibre-fibre distance) distribution before and after the removal of the electrosprayed PEG microparticles (diameter = 17.2 ± 9.4 μm), showing an increasing of 15 and 17% in the number of pores with dimensions between 10 and 20 μm and superior to 20 μm, respectively. Accordingly, the initial number of pores with smaller dimensions (<10 μm) was successfully reduced from 62 to 30%. Expectedly, this upgrading on the fibre-fibre distance could encourage both cell proliferation and infiltration from the TL to the BL [34,35]. In particular for the BL, this pore size distribution will complement the intrinsic advantages of the adopted spiral shaped design (STEP 3 of Fig. 1) since the gaps between the concentric walls (183.7 ± 57.8 μm) of this scaffolding approach are easily capable of not only boosting 3D cell migration, but also guarantee an efficient oxygen/nutrient supply and metabolic waste removal [36,37]. Relatively to the mechanical response, as anticipated, the stress-stain curve (Fig. 2b) presented a linear regime, indicating no considerable buckling, collapsing or densification of the 3D anisotropic fibrous network for the applied compressive regime. The calculated compression modulus was 38.8 ± 2.2 kPa, which is a value compatible with cartilage TE applications considering that it matches the range of values of the native cartilaginous tissue (89.5 ± 48.6 kPa), which presents such considerable variation not only due to its hierarchical anatomy and physiology, but also because of the tissue origin and the conditions of the experiment [4,[38][39][40]. The mechanical tests were performed with swollen PCL scaffolds (Fig. 2c) in order to guarantee the influence of the water uptake capacity during the accommodation period and through the restoration of their original shape after compression, similarly to the natural Fig. 2 Characterisation of the PCL scaffold. a Pore size distribution; b representative stress-strain curve; c swelling ratio response of the cartilaginous ECM [4,5,23]. It is important to notice the success of the presented methodology to build robust bi-layered fibrous systems since neither the mechanical nor the swelling testing have compromised the structural integrity of the anisotropic network, which was able to avoid delamination between the TL and BL and, consequently, maintain the initial geometric and morphologic features.

Cell viability and depth-dependent chondrocyte morphology
The cell viability was analysed via a non-destructive metabolic resazurin assay and, as it is possible to see from Fig. 3a, the PCL scaffold remained biologically viable for the culture period of 21 days. Adding to the excellent levels of biocompatibility, the anisotropic electrospun network proved to be suitable for an efficient cellular proliferation since there was a steadily increasing number of viable chondrocytes over 7 days followed by a plateau that was maintained until the end of the culture. The microscope analysis confirmed that the increased fibre-fibre distance was proficient to promote a successful 3D cell migration across the scaffold, leading, consequently, to the colonisation of the deepest region of the anisotropic system (Fig. 3b, c). In detail, the fluorescence images of the longitudinal cross section of the PCL scaffold showed that the interconnectivity between its two layers generated an efficaciously chondrocyte infiltration from the TL to the BL (Fig. 3d), where the further spreading of the cells and their survival were guaranteed by an optimised nutrient and oxygen supply across the gaps that intercalated the concentric fibrous walls of the spiralled network (Fig. 3e) [41,42]. Furthermore, the distinct effects on the cell morphology induced by the complementary topographic cues of both TL and BL were studied via SEM analysis. The vertical orientation of the electrospun spiral branches relatively to the TL fibrous network (Fig. 4a) was capable of promoting chondrocyte elongation along the fibres with the cells showing lamellipodia and filopodia-like extensions typical of spreading cells (Fig. 4b, c) [43]. Another interesting detail of the 3D migration towards the BL is illustrated in Fig. 4d, where it is possible to observe that the chondrocytes were able to bridge two adjacent fibrous walls of the electrospun spiral in addition to their more frequent columnar arrangement. From the top view of the PCL scaffold (Fig. 4e), it is evidenced not only the contrast between the geometries of both layers, but also the dense cell layer that covers the superficial region. Together with the higher cell density of this zone and contrary to the cells located onto the BL, the chondrocytes attached to the TL of the PCL scaffold showed, as expected, a flattened polygonal morphology characteristic of the 2D electrospun networks (Fig. 4f, g) [44]. This depthdependent morphological dissimilarities between the chondrocytes are summarised in Fig. 4h, i, where it is possible to confirm the biomimetic capability of the BL to induce a more pronounced elongation of the cells comparatively with the TL, as well as the zonal-dependent number of elongated cells (E > 1), which changed from 10% in the superficial region of the PCL scaffold to 59% in the spiralled fibrous section. Overall, the presented biocompatibility results, although preliminary since it is necessary further characterisation of the cell behaviour by evaluating the gene expression and the newly produced ECM, confirm the capability of the PCL scaffold to mimic major depth dependent features of the natural cartilaginous tissue such as cell morphology (flattened and elongated shaped chondrocytes onto the TL and BL, respectively), higher cell density in the superficial region and oriented cell-material interactions (progressing from horizontally aligned in the TL to a vertically arrangement in the BL) [4,45]. Indeed, the anisotropic control of cell density and morphology promoted by this scaffolding approach plausibly complements the advantages reported by other well-established modalities that apply thermal and freeze-drying cycles into the electrospun nanofibres for constructing highly biocompatible 3D PCL fibrous systems, but where cells are homogeneously dispersed through the constructs [46,47].

Conclusion
In this work, we presented a new methodology for recreating in vitro the arcade-like topography of the cartilaginous tissue, where the orientation of the fibres progress from vertical in the deepest zone to horizontal in the superficial zone. Indeed, by sequentially changing the simultaneous electrospinning-electrospraying set up, it was possible to engineer a hierarchical bi-layered PCL scaffold with mechanical properties and pore interconnectivity suitable for cell culture protocols. Additionally, the depth-dependent orientation of the 3D fibrous network was capable of influence the morphology, alignment and density of the cultured chondrocytes likewise its natural counterpart. Thus, it was possible to validate an original and simple microfabrication process able to build 3D biomimetic fibrous architectures compatible with cartilage TE applications. Future work will include a more detailed characterisation of the cell response and its modulation by mechanical stimuli.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.