A composite Gelatin/hyaluronic acid hydrogel as an ECM mimic for developing mesenchymal stem cell‐derived epithelial tissue patches

Here we report fabrication of Gelatin‐based biocomposite films and their application in developing epithelial patches. The films were loaded with an epithelial cell growth factor cocktail and used as an extracellular matrix mimic for in vitro regeneration of organized respiratory epithelium using Calu‐3 cell line and mesenchymal stem cells (MSCs). Our data show differentiation of Calu‐3 cells on composite films as evidenced by tight junction protein expression and barrier formation. The films also supported attachment, migration, and proliferation of alveolar basal epithelial cell line A549. We also show the suitability of the composite films as a biomimetic scaffold and growth factor delivery platform for differentiation of human MSCs to epithelial cells. MSCs differentiation to the epithelial lineage was confirmed by staining for epithelial and stem cell specific markers. Our data show that the MSCs acquire the epithelial characteristics after 2 weeks with significant reduction in vimentin, increase in pan cytokeratin expression, and morphological changes. However, despite the expression of epithelial lineage markers, these cells did not form fully functional tight junctions as evidenced by low expression of junctional protein ZO1. Further optimisation of culture conditions and growth factor cocktail is required to enhance tight junction formation in MSCs‐derived epithelial cells on the composite hydrogels. Nevertheless, our data clearly highlight the possibility of using MSCs in epithelial tissue engineering and the applicability of the composite hydrogels as transferrable extracellular matrix mimics and delivery platforms with potential applications in regenerative medicine and in vitro modelling of barrier tissues.

functional epithelium, in vivo epithelium migration, epithelium interaction with other tissues (immune cells), and effects of underlying tissues on epithelium functionality (Soleas, Paz, Marcus, McGuigan, & Waddell, 2012). Thus far the in vitro development of functional epithelium has not been achieved due to its complex nature, limited differentiation ability of epithelial basal cells and formation of a nonorganized epithelium even when cells are cultured at air-liquid interface (ALI; Vrana et al., 2013). Various studies recommend the potential use of stem cells (embryonic stem cells, induced pluripotency stem cells, and adult stem cells) for epithelium development using biochemical approaches; however, these methods need several cumbersome steps and long culture time to induce the differentiation and thus are unattainable for lab-bench to clinical translation or easy to use as in vitro model systems (Firth et al., 2014;Huang et al., 2014;Ricciardi et al., 2012).
Stem cells are key elements of tissue engineering and offer hope as a therapeutic avenue. However, major consideration when using stem cells for respiratory epithelium development is the identification of the correct stem cells and their capacity for organized epithelial differentiation (Kumar, Vrana, & Ghaemmaghami, 2017). Mesenchymal stem cells (MSCs) are one of the most widely used cells and their presence have been reported in lung tissue and have a role in tissue regeneration in vivo. However, MSCs have shown limited application in in vitro respiratory tissue regeneration. For example, MSCs support the development of respiratory mucosa-like tissue in co-culture with normal human bronchial epithelial (NHBE) but did not acquire the epithelial characteristics (Visage, Dunham, Flint, & Leong, 2004).
The interaction of basement membrane extracellular matrix (ECM) with epithelium stem cells/progenitor cells promotes the airway epithelium repair in vivo by modulating epithelial cell migration and proliferation via differentiation of these progenitor cells to the epithelium subtypes (Coraux, Roux, Jolly, & Birembaut, 2008). The comparative study of lung and bone marrow (BM)-derived MSCs suggests that the epithelial differentiation of BM-MSCs can be achieved in the presence of retinoic acid; however, this effect was minimal compared with lung derived MSCs or epithelium basal cells (Ricciardi et al., 2012). Thus, mesenchymal-to-epithelial-transition is still controversial; however, if successful, this phenomenon can significantly accelerate the respiratory epithelium development using MSCs.
Biomaterial based scaffolds have been used for the development of various tissues (Beckstead et al., 2005;Calejo et al., 2017;Grolik et al., 2012;Tan et al., 2017). However, there is still a need to develop application specific biomaterials with appropriate mechanical properties and capacity to support cell growth, migration, and proliferation. Novel ECM-based delivery platforms carry and deliver the therapeutic agents (e.g., growth factor) in a controlled manner, and simultaneously protect them from fast degradation (Geckil, Xu, Zhang, Moon, & Demirci, 2010).
The mechanical properties of in vitro microenvironment could also direct the stem cell differentiation to a specific cell-lineage (Baeza-Squiban et al., 1994;Engler, Sen, Sweeney, & Discher, 2006;Mendez, Ghaedi, Steinbacher, & Niklason, 2014;Wen et al., 2014). Moreover, the ECM-based scaffolds are potential tools for in vitro tissue development due to their ability to mimic the native microenvironment, for example, collagen type I, collagen IV, laminin, and glycoproteins (Yen, Chan, & Lin, 2010). These ECM components induce epithelial migration via integrin signalling and are thought to play a key role in directing epithelial repair. Thus, the airway epithelial regeneration in the presence of thin biomaterial substrates is considered as one of the possible methods to induce airway epithelium formation for developing robust models which have the basement membrane component (Vrana et al., 2011).
In this context, Gelatin-based biomaterials have been used in supporting the growth of liver (Yan et al., 2005), bone (Yang, Hsu, Wang, Hou, & Lin, 2005), cardiac (Pok, Myers, Madihally, & Jacot, 2013) and skin (Lu, Oh, Kawazoe, Yamagishi, & Chen, 2012) tissues. The mechanical properties of Gelatin can be further manipulated using cross linking and/or encapsulating nanoparticles (Rao, 2007). Controlled cross linking of Gelatinbased scaffolds also induces porous structures, conductive to enhanced cell attachment and differentiation (Yan et al., 2005). The advantage of patch-based delivery is to ensure the positioning of the epithelial cells.
We envision an endoscopic delivery with a releasable clamp that holds the patch in place; with fibroscopy the positioning of the patch can be ensured and once in the correct location the clamp is released to apply the patch to the target surface. As both Gelatin and HA are adhesive by their nature, the establishment of the interface would not be problematic. If necessary, a layer of wet adhesive can be added as we have described previously . Another advantage of the patch system is that the MSCs can continue their differentiation with the right polarity (as the substrate defines their positioning) in a microenvironment that is particularly suitable for respiratory epithelium differentiation.
In this study, we describe fabrication of Gelatin-based biocomposite films loaded with an epithelial cell growth factor cocktail for developing epithelial patches using BM-MSCs. It was hypothesized that growth factor loaded composite films can act as an ECM mimic that is able to facilitate the BM-MSCs differentiation to respiratory epithelium. First, we demonstrated the feasibility of supporting an epithelial layer with a respiratory epithelial cell line (Calu3 cells). Calu-3 is a highly studied respiratory epithelial cell line with the ability to form tight junctions and an efficient barrier in vitro making it suitable for modelling the airway epithelial barrier in respiratory tract research (Grainger, Greenwell, Lockley, Martin, & Forbes, 2006). Moreover, the system was used to assess whether it is capable to support the differentiation of the BM-MSCs towards respiratory epithelium lineage.

| MATERIALS AND METHODS
All tissue culture plastics were purchased from Sarstedt and Nunc. Tissue culture inserts were purchased from Costar Corning. The FGF-7 and FGF-10 were purchased from Peprotech UK. The bronchial epithelium media and stem cell media were purchased from Promocell.
The bronchial differentiation media was prepared using epithelium growth medium from Lonza UK without adding triiodothyronine. All primary and secondary antibodies were purchased from Abcam UK and Thermo Fisher Scientific UK respectively. Gelatin type B (Mw =2-2.5×104 Da, pI = 4.7-5.2) from bovine skin, Fluorescein isothiocyanate labeled bovine Albumin (BSAFITC, Mw = 6.6×104 Da) were (previously installed in the spin coater) and the spin coating program was started. To obtain the self-standing films 60 μl of cellulose acetate (1% w/v in acetone) was deposited on top of the glass slide prior to spin coating process and the sample were dried for 1 day at 4°C. The parameters for spin coating were 2,500 rpm with an acceleration of 1,250 rpm for 2-min time. Afterwards, the films were kept dry at 4°C for at least 3 hr before cross linking. All solutions were prepared using ultrapure water (Milli Q-plus system, Millipore) with a resistivity of 18.2 MΩ.cm and filtered using 0.22-μm filter before use. Again, the non-crosslinked ingredients were washed out using two rinsing steps by incubating with 100 μl of PBS each time for 5 min.

| Thickness determination
The thickness of Gelatin and Gelatin/HA films were estimated with confocal microscope (ZEISS LSM 710). To visualize the film and estimate the thickness, BSA FITC solution (green fluorescent probe) was used to label the films. By reconstructing the whole film thickness using multiple z-stacks can be visualized, hence allowing the determination of the thickness of the film with a 20x objective.

| Film transfer on transwell and growth factors loading
For the biocomposite delivery system, the films were prepared according the protocol described above. A 6% w/v solution of Gelatin type A was prepared in MilliQ water. Gelatin solution was put in water bath at 50°C until complete dissolution. At the same time, a 20% w/v TGA solution was prepared in PBS.

| A549 CELLS EXPERIMENT AND CELLULAR MIGRATION (TIME LAPSE MICROSCOPY)
A549 human lung carcinoma epithelial cells were used as model of respiratory epithelial cells. They were cultured in RPMI 1640 basal Pen-Strep and 1% v/v Fungizone. After trypsinization, 50,000 cells prepared in 30 μl of medium were deposited on top of the Gelatin-HA film previously crosslinked with TGA and HRP and then UV treated for 15 min. After seeding, samples were then left to incubate at 37°C with 5% CO 2 for 7 days. Culture medium was changed every 48 h. Metabolic activity was determined with a resazurin-based test (Sigma Aldrich) and checked at days 1, 3. and 7 to evaluate the proliferation. DAPI/Phalloidin (F-actin) staining was performed after 7 days of culture. After fixation with paraformaldehyde (3.7% v/v in PBS for 15 min), cells were incubated with Triton X-solution (0.1% in PBS for 5 min) and BSA solution (1% v/v in PBS for 20 min). Then samples were incubated for 1 hr with phalloidin (Alexa Fluor 568 phalloidin [6.6μm], Molecular Probes Life Technologies) at a dilution of 1/40 in BSA solution (1% v/v in PBS). After that, two rinsing steps in PBS were performed and the samples were incubated 5 min in Hoechst 33342 solution (20 μg/ml). Finally, samples were visualized with confocal microscope.
For cellular migration analysis, A549 cells were stained with Hoechst 33358 solution (20 μg/ml) for 30 min in a T75 cm 2 for 30 min. Then they were trypsinized and seeded on the different Gelatin-based films (15000 cells/films) for 15 min and mounted in a Ludin Chamber (Life Imaging Services, Basel, Switzerland) at 37°C , 5% CO 2 . Then time-lapse experiment was performed on a Nikon Ti-E microscope equipped with a 10x objective and with an Andor Zyla sCMOS camera and driven by the Nikon NIS-Elements Ar software. Images were acquired every 10 min for 24 hr simultaneously by phase contrast and by fluorescence microscopy with nucleus staining by Hoechst 33342. Cell tracking by the software "NIS-Elements Ar 3D tracking" (Nikon) was carried out in different fields of the substrates. First, the software detected objects "nuclei" by thresholding and they were afterwards tracked over the 24 hr.
Phase contrast images were used to check the viability of the migratory cells. Cells that died during the experiment were eliminated.

| MSCs CELL CULTURE and EPITHELIUM DIFFERENTIATION ON GELATIN-BASED FILM
To get enough number of cells the MSCs (derived from a single donor; Promocell, Germany) were expanded in MSC medium (Promocell Germany). The cells were routinely cultured at 37°C and 5% CO 2 in stem cell media (consist of basal media and medium supplement) as per the manufacturer's protocol. The medium was changed routinely in every 2-3 days up to the confluence. For all experiments the lower passage MSCs (p4-p5) were used.
The MSCs were seeded on Gelatin-based films (50,000 cells per inserts) on a transwell insert in epithelium media (Promocell). The cells were left for 2-3 weeks in submerged culture followed by 2 weeks culture at ALI in epithelium differentiation media (Lonza). For this, the medium from upper chamber was removed and 500-μl differentiation medium was used in lower chamber. The experiment was also repeated using DMEM-F12 media to assess the impact of the Gelatin films in the absence of growth factor in the differentiation media.
Moreover, to assess the impact of Rho-associated protein kinase (ROCK) inhibition on MSCs differentiation towards respiratory epithelium; 1 μl/ml of Rho kinase inhibitor Y-27632 was added throughout the culture in a separate experiment. In each case, the medium was changed every 2-3 days.

| CALU-3 CELL SEEDING AND DIFFERENTIATION ON GELATINE-BASED FILMS
Calu-3 cells were used as a model epithelial cell to assess barrier formation. Briefly, Calu-3 cells were seeded on Gel-HA film (within transwell) at 50,000 cells/CM 2 in DMEM-F12 medium with 10% FBS, 1% P/S, 100 mM non-essential amino acid and 100 mM of L glutamine supplement. The cells were transferred to ALI after growing in submerged conditions for a week.

| Immunofluorescence for epithelium differentiation
After submerged culture of MSCs for 2-3 weeks the expression of epithelium and MSC markers was assessed using various epithelium markers; wide spectrum cytokeratin, pan cytokeratin and cytokeratin 18 (1:100 dilution) and mesenchymal marker; vimentin (1:200 dilution) antibodies (all antibodies from abcam). At the end of each culture, the cells were fixed with 3% formaldehyde in PBS (pH 7.4) for 30 min and were permeabilized using the 0.25% Triton X 100 solution for 20 min.
After each step, the samples were washed three times in PBS (5 min).
The samples were further incubated with 3% BSA for 1 hr to stop the non-specific binding of proteins. Cell layers were incubated with primary antibodies diluted in PBS for 90 min at room temperature. After primary incubation, the samples were incubated with secondary antibody for 45 min at room temperature. The secondary antibodies used were Alexa Fluor® 488, chicken anti rabbit 1:250 dilution in PBS or rhodamine red anti-mouse 1:250 dilution in PBS). Each antibody incubation was followed by three washes in PBS for 5 min. For nuclear staining, DAPI (4′ 6-diamidino-2-phenylindole) was used at 1: 4000 dilution (Invitrogen, UK) for 15 min and followed by 3 PBS washing for 5 min. Finally, the samples were mounted on glass slides with VectaShield (Vector Laboratories, UK) for direct observation. The images were taken using Leica DMRB fluorescence microscope using 10X or 40X objective. The intensity of the fluorescence images was evaluated using the image J software.

| Trans-epithelial electrical resistance measurements
The trans-epithelial electrical resistance (TEER) measurements across the cell monolayer cultured at the ALI were performed according to the method described elsewhere (Harrington et al., 2014). Briefly; the measurements were performed using an EVOM volt-ohm-meter and STX2 chopstick electrodes (World Precision Instruments, U.K.).
The cell culture media was added to the upper chamber (500 μl) and lower chamber (1.5-ml total volume) and allowed to equilibrate for 30 min prior to measurements (37°C, 5% CO 2 ). The TEER values were measured several times in several samples. The chopstick electrodes were sterilized using 70% v/v ethanol in distilled water.
Control measurements were performed using Calu 3 cell lines cultured in a similar way.

| ZO-1 and Mucin staining
At the end of ALI culture, the samples were fixed with 3% formalde-

| STATISTICAL ANALYSIS
The statistical analysis was performed using GraphPad Prism version 6.00 for Windows, GraphPad Software, La Jolla, California USA (www.graphpad.com). All results are shown as mean ± standard deviation (SD) from three independent experiments. Statistical differences were determined using the student t-test or one-way analysis of variance method with Tukey post-hoc testing. A p-value <0.05 was considered statistically significant.

| Composite film fabrication and their protein release profile
The epithelium stem cells are scarce and have limited regenerative capacity for epithelium tissue regeneration thus there is an imperative need for alternative cell sources which can potentially be used for airway epithelium development (Vrana et al., 2013). The current literature suggests the significant contribution of various stem cells in epithelium regeneration, for example, MSCs, ESCs, and iPSCs. The MSCs have been widely used for the development of bone, cartilage and adipose tissue, originated from the mesodermal layer of embryo. Recent literature also provides some evidence for MSCs differentiation towards non-mesodermal cell and tissue development, for example, neuronal, liver, epithelium and pancreatic tissue. Some recent studies also suggested the differentiation capability of BM-MSCs to epithelial cells in vitro (Păunescu et al., 2007) and in vivo (Kotton et al., 2001). They acquire phenotypic and functional epithelial characteristics when cocultured with airway epithelial cells (Ma et al., 2011;Spees et al., 2003). Providing appropriate growth factors and mechanical and biochemical cues that simulate ECM in the basal membrane could potentially support MSCs differentiation towards epithelial cells, removing the need for complex co-cultures. Thus, we proposed to use a Gelatinbased bio-composite film capable of controlled release of epithelium inducing growth factors (FGF-7 + FGF-10) as a ECM mimic for developing a respiratory epithelium patch using BM-MSCs.
Gelatin, as a natural biomaterial, has been used in various tissue regeneration applications in two-dimensional and three-dimensional cross-linked form (Sell et al., 2010). The physiochemical properties of the Gelatin can be controlled easily for development of substrates for cells, for example, the controlled cross linking of Gelatin-based scaffold induces porous structure, conductive to enhanced cell attachment and differentiation (Yan et al., 2005) In the first part of this study, the fabrication method of Gelatin film was optimized to determine the main parameters that will influence the thickness of the film. To do that, different Gelatin concentrations and rotation speeds were tested to spin coat and the thickness of the resulting film after crosslinking with TGA was estimated with confocal microscope after loading of BSA FITC to visualize the cross section ( Figure 1a). Our data clearly show that the main parameter that influences the thickness of the film is Gelatin concentration. The difference in gel thickness between three different rotation speeds for the same concentration was not significant. The only trend observed was the increase in the thickness with Gelatin concentration. To have better stability and thicker film, we worked with 15% w/v Gelatin concentration and used a rotation speed of 2500 rpm for the rest of the study.
To better simulate the composition of the basement membrane in vivo, we incorporate HA in the film formulation. In a previous work from our lab (Knopf-Marques et al., 2017), we have demonstrated that the stability of Gelatin/HA membrane films can be improved using HA derivative such as HA-tyramine by creating an interpenetrated network through a double crosslinking step. Gelatin is crosslinked with TGA to create amide bond between amine groups on lysine residues and carboxamide groups on glutamine residues and HA-tyramine is crosslinked through the formation of dityramine groups in the presence of HRP (HRP). HA-tyramine was incorporated in the film formulation with the following ratio (Gelatin 14%/HA-tyramine 1% w/v) and referred to as Gelatin-HA. The addition of HA in the structure resulted in an increase in the film thickness, from about 10μm for Gelatin to 15.5μm for Gelatin/HA (Figure 1b). This difference in thickness can be attributed to the intrinsic capacity of HA to absorb large amount of water.
As these Gelatin-based membranes are supposed to release growth factors for the differentiation of MSCs to epithelial cells, the release property of these materials was tested using a fluorescently labelled model protein BSA FITC . The release property of Gelatin membrane was tested in a previous work from our group (Barthes et al., 2015) and we reproduced this experiment with Gelatin/HA film and the cumulative release of BSA FITC was followed at 37°C in PBS solution by quantifying the fluorescence in the supernatant using a spectrofluorimeter (Figure 1c). Both materials have shown the ability for the loading and the release of bioactive molecules for at least 2 weeks after an initial burst release.

| The composite bio-film supports migration, growth, and differentiation of alveolar epithelial cells
The next step was to compare the behaviour of both film formulations (Gelatin vs Gelatin-HA) on A549 epithelial cells in terms of migration distance and migration speed to see if the addition of HA in the formulation had an effect. A549 cells were selected due to their aggressive migratory nature (Ciftci et al., 2016) with the hypothesis that with a more migratory cell type the effect of the substrate will be accentuated and easier to quantify. This experiment was carried out using Time Lapse microscopy for 24 hr. It was shown that the addition of HA in the formulation did not have an effect on both cell migration distance and cell migration speed. Both materials exhibited the same response toward A549 epithelial cells (Figures 2a and 2b). Moreover EGF (epidermal growth factor) was also loaded in Gelatin-HA film and the same experiment of migration was repeated. The presence of this growth factor also did not affect cell migration and cell migration distance when we compare to both Gelatin and Gelatin-HA film. The proliferation of A549 epithelial cells on Gelatin material has been studied in a previous work from our group (Ciftci et al., 2016) and we have repeated the same experiment with Gelatin-HA film (Figure 2c). The metabolic activity was followed for 7 days and a significant increase was seen between days 1, 3, and 7 meaning that cells were proliferating on the film.

| MSCs growth and differentiation to epithelial like cells on Gelatine-based films
The MSCs were seeded on Gelatin and Gelatin-HA films and cultured in submerged cultures for up to 3 weeks to adapt to epithelium environment using epithelium media. It is evident that the MSCs seeded on various films adhered well on films and were viable even up to 3-4 weeks of culture. The epithelium cells are cobblestone shaped cells smaller in size than the mesenchymal cells. The morphology of BM-MSCs seeded on transwell or Gelatin-based films changed into round shape morphology, similar to epithelium. However, the stem cells in standard medium still demonstrate the spindle morphology ( Figure 3a). The cells on Gelatin-based film without the growth factors also demonstrate the smaller size due to the emulation of softer substrate similar to native epithelium ECM and thus the physicochemical and mechanical properties of culture microenvironment could control the MSC cell behaviour (Li et al., 2010). The stiffness of dense transwell membranes are significantly higher than the mechanical properties of the described films that have been previously reported to be between 30 and 50 kPas (Knopf-Marques et al., 2017). Previous work on uncrosslinked Matrigel showed a Young's modulus of 440±250 Pa (Soofi, Last, Liliensiek, Nealey, & Murphy, 2009)  the presence of growth factor enriched medium (Păunescu et al., 2007). However, the authors did not report the formation of intercellular tight junctions.

| Impact of ROCK inhibition on MSCs full differentiation to epithelial cells
In order to enhance the differentiation of MSCs to respiratory epithelium we investigated the impact of ROCK inhibition. ROCK inhibition using Rho kinase inhibitors like Y-27632 has been shown to promote the induction efficiency, self-renewal, and differentiation of MSCs towards the keratinocyte like cells with good expression of cytokeratin 14 and cytokeratin 5 in the presence keratinocyteconditioned medium (Li et al., 2015). Thus, it was envisaged that this approach can also further push the epithelial differentiation of MSCs in the presence of biocomposite films. Accordingly, we performed

| MSCs as a potential cell source for developing personalized epithelial patches
Given their more accessible nature and availability in larger numbers (e.g., compared with epithelial cells from nasal turbinate), MSCs could be potentially used for development of personalized epithelial models.
The provision of an ECM like membrane layer with appropriate mechanical properties further provides a means to mimic the effects of ECM membrane on the behaviour of healthy and diseased epithelial cells. However, for a more robust model further development that would induce better barrier function and cell-cell contact is required.
In order to see if the developed composite ECM membrane can be used for more general respiratory epithelium models, we also cultured a well-established epithelial cell line, Calu 3 ( Figure 6) on these films.
Staining for Mucin and ZO-1 demonstrated mucin secretion and tight junction formation by the cells on the surface of ECM mimics. This further highlights the potential application of the biocomposite films for developing respiratory epithelium models.

| CONCLUSION
Using MSC-derived epithelial cells would provide a larger cell pool (compared with primary respiratory epithelium) and patient specificity (compared with respiratory epithelium cell lines) that could enable development of more personalized tissue models with clear benefits for disease modelling or testing new drug leads. Such endeavour also requires ECM mimicking substrates that are adaptable to air liquid interface culture conditions to provide more physiological relevance.
Herein, we describe composite film that mimics the ECM components and basal membrane with high stability for long term culture periods and the capacity to release growth factors under ALI conditions. MSCs cultured on these substrates in the presence of growth factors showed substantial decrease in mesenchymal marker expression and increased epithelial marker expression. ROCK inhibition provided a more advanced differentiation. This study demonstrates the feasibility of using growth factor loaded biocomposite films and MSCs for development of in vitro respiratory epithelial models. Our future work will focus on optimisation of culture conditions including modifying the composition of the films and GF delivery conditions to induce better mucin secretion and epithelial barrier function.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.    GFs loaded Gelatin-HA film. The TEER measurement further confirmed the formation of intercellular tight junctions. However the effects were not significant more than GFs loaded film.