Non‐woven bilayered biodegradable chitosan‐gelatin‐polylactide scaffold for bioengineering of tracheal epithelium

Abstract Objectives The conversion of tissue engineering into a routine clinical tool cannot be achieved without a deep understanding of the interaction between cells and scaffolds during the process of tissue formation in an artificial environment. Here, we have investigated the cultivation conditions and structural features of the biodegradable non‐woven material in order to obtain a well‐differentiated human airway epithelium. Materials and methods The bilayered scaffold was fabricated by electrospinning technology. The efficiency of the scaffold has been evaluated using MTT cell proliferation assay, histology, immunofluorescence and electron microscopy. Results With the use of a copolymer of chitosan‐gelatin‐poly‐l‐lactide, a bilayered non‐woven scaffold was generated and characterized. The optimal structural parameters of both layers for cell proliferation and differentiation were determined. The basal airway epithelial cells differentiated into ciliary and goblet cells and formed pseudostratified epithelial layer on the surface of the scaffold. In addition, keratinocytes formed a skin equivalent when seeded on the same scaffold. A comparative analysis of growth and differentiation for both types of epithelium was performed. Conclusions The structural parameters of nanofibres should be selected experimentally depending on polymer composition. The major challenges on the way to obtain the well‐differentiated equivalent of respiratory epithelium on non‐woven scaffold include the following: the balance between scaffold permeability and thickness, proper combination of synthetic and natural components, and culture conditions sufficient for co‐culturing of airway epithelial cells and fibroblasts. For generation of skin equivalent, the lack of diffusion is not so critical as for pseudostratified airway epithelium.


| INTRODUC TI ON
The factual knowledge, accumulated by tissue engineering so far, is not sufficient for a transition from the empirical to a fundamentally new, rationale-based level. Without a thorough understanding of mechanisms, that drive tissue formation in an artificial environment and the interaction of cultured cells and biodegradable scaffolds, one could not expect any significant progress in the field.
In epithelia bioengineering, biodegradable materials are of primary interest, since the restoration of the barrier function requires a timely replacement of an artificial scaffold with a native tissue, thus preventing the infection and excessive scarring. 1 The process of formation and maturation of the functional epithelium requires reciprocal interaction of epithelial and mesenchymal cells. Two types of bilayered scaffolds for the 3D cultivation of respiratory epithelium deserve attention at this stage. The first type represents a bilayered matrix combining a film-like top layer and porous sub-layer (a collagen-or collagen/hyaluronate-based sponge).
Although such collagen-based materials are sufficient in terms of biocompatibility and biodegradability, they are not providing adequate mechanical properties. 6 Another material, proposed by Morris and coworkers, 7 also represents a bilayered but non-woven scaffold made from a non-biodegradable polyethylene terephthalate. This scaffold mimics the fibrous structure of a natural extracellular matrix (ECM) of the decellularized trachea and was shown to be suitable for co-cultivation of fibroblasts (in the microfibrous layer) and lung adenocarcinoma CALU3 cells (placed on top of a nanofibrous layer) in a bioreactor. 8 Surprisingly, while two types of scaffold discussed above are probably the best of their kind, a solid proof that they are sufficient to generate functional equivalents of the airway epithelium is still lacking. So far no data have been reported on the developing of differentiated airway epithelium with primary human cells on non-woven scaffolds.
Here, we obtained a novel mechanically sound polymeric scaffold that combines the advantages of both types of scaffolds mentioned above. We investigated the structural features of the biodegradable non-woven material and the cultivation conditions to obtain a well-differentiated human airway epithelium equivalent. On the same scaffold, we obtained skin equivalent with keratinocytes and analysed growth and differentiation for both types of epithelium. The scaffold was identified as a prospective tool for bioengineering of epithelial tissues including both airway epithelium and skin.

| Scaffold fabrication
Copolymer films were cast from a stable colloidal solution of chitosan-gelatin-poly-l-lactide copolymer (CGP) in dichloromethane (DCM; Component-Reactiv, Russia) as previously described. 10 A bilayered fibrous matrix was obtained from CGP polymer dispersions in DCM and DCM:ethyl alcohol (Himzakaz, Russia) using electrospinning technique with equipment previously described. 11 Addition of PLLA (4032D; NatureWorks, USA) to CGP dispersion (53% CGP + 47% PLLA) allowed to obtain a matrix with the desired mechanical properties. The parameters of a time-stable spinning process are described in Table 1. Three types of matrices with nanolayer 25, 50 and 100 μm thick were obtained.

| Surface modification
Direct current discharge plasma modification (surface activation) was carried as described earlier. 10

| Pore size analysis
The average pore size of the scaffolds was calculated in triplicate using POROLUX™ 1000 capillary flow porometer (Porotec, Germany) according to the manufacturer's protocol.

| Mechanical properties of scaffold samples
The mechanical properties of matrices were determined in triplicate

| Two-chamber cell culture system
For the cultivation at air-liquid interface conditions (ALI), a customized culture system was created similar to the one previously described. 5 Polymer membranes were removed from 12-mm-diameter cell culture inserts (Millicell, Cat. No. PIHP01250 and Corning, Cat. No. 3460), and the matrix was clamped between their plastic frames.
The construct was placed into a 12-well plate.

| Cultivation of fibroblasts to evaluate their effects upon mechanical characteristics of the matrix
For mechanical testing, fibroblasts were seeded onto the microfibrous surface of the scaffold and cultivated as described below in Section 2.4.3 for 14 or 30 days.

| Development of complex (multicellular) equivalents
The matrix was pre-moistened in DMEM at 37°C for 1 hour be- Each type of equivalent was obtained at least three times with cells from different donors every time.

| Development of monoculture epithelial equivalents
The epithelial cells were seeded on pre-soaked matrices and cultivated as described above. For positive control, 12-mm-diameter cell culture inserts (Millicell, Cat. No. PIHP01250) were used along with CGP matrices.

| Cell viability analysis
Evaluation of the proliferation was carried out using the MTT cell

| Histological analysis
The equivalents were fixed with 1.5% glutaraldehyde and 2% osmium tetroxide (SPI-Chem, Cat.No. 02601-AB), followed by dehydration through a graded series of ethanol and embedded in

| Scanning electron microscopy
Cell-seeded scaffolds were fixed overnight in 2.5% glutaraldehyde, followed by dehydration through a graded series of ethanol and dried on a critical point dryer HCP-2 (Hitachi Company, Japan). Sputtering of gold on the sample was performed with IB-3 Ion Coater (Eiko, Japan). Samples were examined under a scanning electron microscope Vega TC CamScan MV2300 (CamScan, UK).

| Statistical analysis
The results were presented as a mean ± standard error. Statistical analysis was performed using the Student's t test. At P-values <0.05, the results were considered as statistically significant.

| Fabrication and modification of a bilayered material
The morphology and structural characteristics of the obtained scaffolds are shown on Figure 1 and Table 2, respectively. Both nano-and microfibrous layers were homogenous, free of beads and possessed an anticipated structure. The upper (nanofibrous) layer ( Figure 1A) was designed to support the growth of epithelial cells and to prevent their migration into interior space of the scaffold. The inter-fibre distance (pore size) in this layer was 2.8 ± 0.1 μm, which corresponded to a fibre diameter of 0.35 ± 0.2 μm ( Figure 1C,E and Table 2). The lower microfibrous layer was designed to allow the fibroblasts to migrate inside and evenly populate the entire volume of the layer. This layer had an inter-fibre distance of 37.5 ± 3.8 μm, which corresponded to a fibre diameter of 2.5 ± 2.4 μm ( Figure 1D,F and Table 2). The thickness of microfibrous layer was always the same (180 ± 3 μm), while for nanolayer, we tested three different options: 25, 53 and 100 μm. Both layers were tightly and securely connected to each other ( Figure 1B).
After the electrospinning, the scaffold surface was hydrophobic (the water contact angle was 114°). To increase hydrophilicity of the material, the scaffold was subjected to modification comprising plasma treatment and subsequent HA immobilization. HA treatment made the measurement of the water contact angle impossible since the water was instantly absorbed into the material (Video S1).

| Proliferation of epithelial cells depends on nanofibrous surface structure at submerged culture
The fluorescence microscopy and SEM have shown that the primary human respiratory epithelium cells actively proliferate, migrate and reach confluence on the surface of the nanofibrous layer with practically no dead cells as confirmed by viability/cytotoxicity assay ( Figure 2A). On days 2-4 of culturing, the cells covered the entire surface of the matrix and formed multiple intercellular contacts, as indicated by staining with antibodies to claudin I (data not shown) and β-catenin ( Figure 2D), as well as by SEM results (Figure 2B,C). Most of the cells expressed cytokeratin 5 (CK5)-a marker of basal epithelial cells ( Figure 2E). This indicates that cells retain the proliferative potential during the cultivation on the matrix and confirms their epithelial origin. Interestingly, if the pore size in the nanolayer was increased to more than 5 μm, the cells tended to form clusters but not a monolayer.
On nanolayer with pores >10 μm, the epithelial cells ceased proliferation and spreading and eventually died ( Figure 2F). Uneven structure of nanofibres (ie, presence of beads) also hampered monolayer formation.
The thickness of the nanolayer also had a prominent effect on the proliferation and migration of epithelial cells (Figure 3). To reach a compromise between the rates of culture medium diffusion through the nanolayer during ALI cultivation and ability of this layer to provide adequate support for epithelialization, we have selected for further experiments the nanolayer of medium thickness of 53 ± 5 μm ( Figure 1B and Table 2). The nanolayer of 25μm thickness appeared to be too rarefied to support cell migration, while 100 μm layer was too thick to provide sufficient nutrient diffusion to the air-exposed cells on its surface.

| Differentiation of epithelial cells at ALI
When cultivated on a non-modified matrix (total thickness of 235 µm, no HA treatment), the epithelial cells survived and proliferated, but the respiratory epithelium was not formed ( Figure

| Positive control (cell cultivation on polycarbonate inserts)
We used commercial polycarbonate inserts coated with collagen IV as a positive control model. The formation of pseudostratified epithelium was confirmed by microscopy ( Figure S1).

| Co-cultivation of epitheliocytes with fibroblasts did not lead to increase in cell differentiation rates
The microfibrous layer retained mechanical properties ( Figure S2) and supported the growth of tracheal fibroblasts and gingival MSCs ( Figure S3). However, neither of these cell types was capable of stimulating epithelial cell differentiation upon co-cultivation. Furthermore, in some 3D cultivation experiments, CK5 positivity was observed all over the epithelium, suggesting stimulation of proliferative response in basal cells and suppression of differentiation ( Figure S3). Surprisingly, in some cases, epithelial cells even died.

| The formation of skin equivalents on bilayered scaffold at ALI
We assumed that the combination of high biocompatibility of CGPbased scaffold with its bilayered structure would support the formation of not only respiratory, but other types of epithelia as well. We evaluated the ability of our material to support growth of 3D skin equivalents using both primary human keratinocytes and dermal fibroblasts.
The microfibrous layer supported growth and migration of dermal Keratinocytes of the basal layer synthesized the component of the basement membrane, collagen IV ( Figure 5D,E). However, analysis of the morphology showed that epithelium was unevenly distributed on the matrix surface: the epithelial strata were thinner towards the centre ( Figure 5F,G). Immobilization of HA resulted in more uniform epithelium morphology in the centre of skin equivalent.
As a positive control, commercial polycarbonate inserts were used ( Figure S4).
Currently, cultivation of respiratory epithelial equivalents is usually performed on commercial polycarbonate membranes 19 or on scaffolds made of collagen 20 or of its mixtures with natural ECM components. 21,22 However, the use of collagen-based scaffolds is hampered by the insufficient mechanical properties of these materials. 6 In addition, while being nearly ideal for cell growth support, they have poor suitability for electrospinning and excessively high biodegradation rates. 23 To tackle these conflicting demands, we have chosen the CGP copolymer combining high mechanical properties and suitability for electrospinning of polylactide 9,24 with biocompatibility, high cell adhesion capacity and antibacterial properties of chitosan. 25 The inclusion of gelatin into CGP backbone provides natural determinants. 26 The combination of acidic (PLLA) and basic (chitosan) polymers in one

| Proliferation of epithelial cells depends on nanofibrous surface structure at submerged culture
According to our results, in bilayered scaffold, the nanolayer should have small pores, be homogeneous and dense (with tightly packed fibres). The epithelial cells do not survive on a nanolayer with irregular structure and multiple beads on the fibres. The appearance of beads was directly related to the ratio of the hydrophobic to hydrophilic components in the polymer dispersion. Supplementation of the final blend with PLLA allowed to obtain more homogeneous fibres and to minimize formation of beads in the nanolayer.
The nanolayer pore size up to 5 μm appeared to be optimal for the epithelial basal cell survival and preservation of their proliferative and migratory potential for sufficient period of time ( Figure 2E).
When a pore size was more than 5 μm, cells formed clusters and did not survive after transfer to ALI.
The nanolayer thickness appeared to be of particular importance for the epithelial cell survival on top of the scaffold. In the submerged culture, 100 μm thick nanolayer supported cell migration and formation of a cell sheath much better as compared to 50 and, especially, 25 μm thick layers. However, during ALI cultivation, when the importance of nutrient supply from the below came to the first place, the 100 μm nanolayer was too thick to provide sufficient diffusion. Thus, for epithelial cell monolayer formation at ALI, the nanofibrous layer of the intermediate thickness (about 50 μm) appears to be optimal.

| Modification of CGP scaffold with hyaluronic acid
The CGP copolymer contained only a part of chitosan and gelatin, and thus cannot be designated as "natural." To gain better control over cell differentiation on CGP matrices, they were modified with

| The role of matrix thickness and permeability
During ALI cultivation, respiratory epithelium cells appeared to be highly dependent upon nutrient and liquid supply from the bottom.
High nutrient demand of developing respiratory epithelium may be linked to high energy consumption by mucociliary transport. To fulfil these requirements, the cultivation of respiratory epithelium equivalents in perfusion bioreactors may represent a good option.
It was previously reported that for artificial scaffolds thicker than 500 μm, forceful perfusion is mandatory. 29 The perfusion leads to a measurable increase in the proliferation rate of oral mucosa cells on sponges. 30 Of note, on thick scaffolds (over 300 µm), we did not see formation of mucociliary pseudostratified epithelium even after HA treatment. The differentiation slipped towards stratified squamous epithelium ( Figure 4M-P). Previously, it was shown that the differentiation of the tracheal epithelium depends on the porosity of the scaffold used. 31,32 Our results are in line with these reports.

| Co-cultivation of airway epithelial cells with fibroblasts
The differentiation of the respiratory epithelium is a complex process controlled by multiple factors, 33 which significantly vary between different mammalian species. 32 Many of these factors are directly secreted by resident fibroblasts or are under fibroblast expressional control. 33 The stimulation of rat and guinea pig tra- When cultured at ALI, the epidermal keratinocytes appeared to be less sensitive to scaffold thickness and porosity as compared to respiratory epithelial cells. With increasing scaffold thickness, the differentiation of airway epithelial cells first shifted towards the flattening squamous type, and then (on thicker matrices) the cells died. In contrast, the epidermal keratinocytes survived on excessively thick (>500 μm) scaffolds and formed stratified layers but the basal layer showed signs of dystrophy. These diversities may be linked to differences in nutrient and moisturization demand between ciliated mucous-producing respiratory epithelium and cornifying epidermis.
Furthermore, for airway epithelium, modification of the CGP scaffold by HA was mandatory to induce formation of columnar epithelium, while in skin equivalents, HA was not essential for differentiation.

| CON CLUS IONS
1. Non-woven materials appear to be extremely attractive for the production of complex epithelial equivalents. However, the production of high-quality non-woven scaffolds requires the use of a high proportion of synthetic components. This makes it difficult to develop mucous airway epithelial equivalents at such non-physiological conditions.

2.
The structural parameters of nanofibres should be selected experimentally depending on polymer composition and changes in surface topography due to scaffold biodegradation. For CGP copolymer used in our studies, optimal thickness of nanofibrous layer is 50 µm (less was too rarefied but more was not enough for diffusion) with pore size not more than 5 µm.

3.
For a stimulating effect of fibroblasts on mucociliary differentiation of epithelial cells, proper cultivation conditions and correct mesenchymal cell source selection are of primary importance.

4.
For the generation of skin equivalent, the lack of diffusion is not as critical as for pseudostratified airway epithelium. This finding may explain why attempts to create skin equivalents on different matrices are much more successful as compared with airway epithelium.
The use of forceful perfusion (a bioreactor) may solve this problem.

ACK N OWLED G EM ENTS
The authors are grateful to Dr. M.S. Piskarev (ISPM RAS) for plasma treatment of polymeric scaffolds and for contact angles measurements, and to Dr. A.L. Vasiliev for assistance in obtaining of SEM images. The study was carried out using equipment of the Resource Centers of Kurchatov Complex of NBICS-Technologies.

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

AUTH O R CO NTR I B UTI O N S
OAR conceived and designed the study. TSD and THT selected and